Tecnico Teiko (tecฺsuporte@teikoฺcomฺbr) has a non-transferable license to use this Student Guideฺ

Unauthorized reproduction or distribution prohibitedฺ Copyright© 2013, Oracle and/or its affiliatesฺ
le
fe rab
tra ns
n -
a
Exadata Database no
Machine
a s
h eฺ Workshop
ฺ b r )
Administrationid
o m G u
oฺc uden t
i k
te St
r @
te this Student Guide - Volume II
o
up use
ฺ s
( t ec se to
e i ko licen
o T
ic
Tecn
D73668GC11
Edition 1.1
April 2012
D76623
Author Copyright © 2012, Oracle and/or its affiliates. All rights reserved.
Peter Fusek Disclaimer
This document contains proprietary information and is protected by

Technical Contributors and Reviewers copyright and other intellectual property laws. You may copy and print
this document solely for your own use in an Oracle training course. The
Akshay Shah Nilesh Choudhury document may not be modified or altered in any way. Except where
your use constitutes "fair use" under copyright law, you may not use,
Alex Tsukerman Ravindra Dani
share, download, upload, copy, print, display, perform, reproduce,

Amit Ganesh Raymond Dutcher publish, license, post, transmit, or distribute this document in whole or
Andrew Babb Richard Exley in part without the express authorization of Oracle.
Aslam Edah-Tally Robert Carlin The information contained in this document is subject to change without
Barb Lundhild Robert Pastijn notice. If you find any problems in the document, please report them in
Roger Hansen writing to: Oracle University, 500 Oracle Parkway, Redwood Shores,
Bharat Baddepudi California 94065 USA. This document is not warranted to be error-free.
Bill Hodak Sabyasachi Banerjee
Boris Erlikhman Scott Gossett Restricted Rights Notice
Branislav Valny Sriram Palapudi If this documentation is delivered to the United States Government or
le
Bruce Kyro Steven Lemme
Sue Lee Government, the following notice is applicable:
fe rab
anyone using the documentation on behalf of the United States
ns
Caroline Johnston
Christian Craft Sugam Pandey U.S. GOVERNMENT RIGHTS
- tr a
Dan Norris Sumeet Lahorani
no n
The U.S. Government’s rights to use, modify, reproduce, release,
perform, display, or disclose these training materials are restricted by
Dave Winter Sundararaman Sridharan
s a
the terms of the applicable Oracle license agreement and/or the
David Hitchcock Tim Shelter a
h eฺ
applicable U.S. Government contract.
)
Douglas Utzig Umesh Panchaksharaiah
b r
ฺ Guid
Ed Gilowski Uwe Hesse m
Trademark Notice
Eric Siglin Varun Malhotra

k o ฺco dent
Oracle and Java are registered trademarks of Oracle and/or its
Georg Schmidt Vern Wagman
t ei Stuaffiliates. Other names may be trademarks of their respective owners.
Harald van Breederode Vijay Sridharan

Vikram Kapooro r te@ this
up use
James He
Jean-Francois Verrier
ฺ s
Vimala Jacob
c e to
Jia Shi
( t eGraphic
Jim Hall
e i ko licens Designer
Jim Spiller
o T Satish Bettegowda
ic
Jim Viscusi
ecn
Joel Goodman
T
Juan Loaiza
Editors
Malavika Jinka
Kam Shergill Smita Kommini
Kevin Jernigan Aju Kumar
Kodi Umamageswaran
Krishnanjani Chitta Publisher
Larry Justice
Michael Sebastian Almeida
Lawrence To
Louis Nagode
Mahesh Subramaniam
Maria Billings
Mark Fuller
Mark Scardina
Mark Van de Wiel
Marshall Presser
Martin Jensen
Michael Cebulla
Michael Nowak
Contents
1 Introduction
Course Objectives 1-2
Audience and Prerequisites 1-3
Course Contents 1-4
Terminology 1-5
Additional Resources 1-6
le
Practice 1 Overview: Introducing the Laboratory Environment 1-7
fe rab
2 Exadata Database Machine Overview tra ns
n -
Objectives 2-2
a no
Introducing Database Machine 2-3 a s
h eฺ
Why Database Machine? 2-4 r )
ฺ Guid
b
Introducing Exadata Storage Server 2-6m
o ฺco dent
Exadata Storage Server Architecture Overview 2-7
k
t ei Stu
Exadata Storage Server Features Overview 2-8
r te@ this
Exadata Storage Server X2-2 Hardware Details (Sun Fire X4270 M2) 2-10
o
s up use
Exadata Storage Server X2-2 Specifications 2-11
ฺ
t ec se to
Database Machine X2-2 Full Rack 2-12
(
i ko licen
X2-2 Database Server Hardware Details (Sun Fire X4170 M2) 2-13
e
o T
Start Small and Grow 2-14
ic
Tecn Database Machine X2-8 Full Rack 2-15
X2-8 Database Server Hardware Details (Sun Fire X4800 M2) 2-16
Database Machine Capacity 2-17
Database Machine Performance 2-18
Exadata Storage Expansion Racks 2-19
InfiniBand Network 2-20
Database Machine Support Overview 2-21
Database Machine Benefits for Data Warehousing 2-22
Database Machine Benefits for OLTP 2-24
Quiz 2-25
Summary 2-27
iii
3 Exadata Database Machine Architecture
Objectives 3-2
Database Machine Architecture Overview 3-3
Database Machine Network Architecture 3-5
InfiniBand Network Architecture 3-7
X2-2 Full Rack Leaf Switch Topology 3-8

Spine and Leaf Topology 3-9
Scale Performance and Capacity 3-10
Typical Scaling Scenarios 3-11
Scaling Out to Eight Racks 3-13
Scaling Out Beyond Eight Racks 3-14
Interconnecting Quarter Racks 3-15 le
InfiniBand Network External Connectivity 3-17 fe rab
Database Machine Software Architecture Overview 3-18
tra ns
n -
Database Machine Software Architecture Details 3-21
Disk Storage Entities and Relationships 3-22 a no
a s
h eฺ
Flash Storage Entities and Relationships 3-24 )
Disk Group Configuration 3-25 b r
ฺ Guid
m
Quiz 3-26
k o ฺco dent
Summary 3-30 t ei Stu
o r te@ this
ฺ s up use
Practice 3 Overview: Introducing Exadata Cell Architecture 3-32
( t ec se to
4 ko liceofn Exadata Database Machine
Key Capabilities
e i
T
Objectives 4-2
o Classic Database I/O and SQL Processing Model
ic
ecn
4-3
T Exadata Smart Scan Model 4-4
Exadata Smart Storage Capabilities 4-5
Exadata Smart Scan Scale-Out Example 4-8
Exadata Hybrid Columnar Compression Overview 4-11
Exadata Hybrid Columnar Compression Data Organization 4-12
Exadata Smart Flash Cache Overview 4-13
Exadata Smart Flash Cache Intelligent Caching 4-14
Using Exadata Smart Flash Cache 4-15
Exadata Smart Flash Log Overview 4-17
Exadata Storage Index Overview 4-18
Storage Index with Partitions Example 4-20
Database File System 4-21
I/O Resource Management Overview 4-22
Benefits Multiply 4-23
Quiz 4-24
iv
Summary 4-25
Practice 4 Overview: Introducing Exadata Features 4-28
5 Exadata Database Machine Initial Configuration

Objectives 5-2
Database Machine Implementation Overview 5-3
Key Documents 5-5
Database Machine Site Preparation 5-6
Configuration Worksheets Overview 5-7
General Configuration Worksheet 5-8
Choosing the Right Protection Level 5-10 le
General Network Configuration Worksheet 5-12 fe rab
Management Network Configuration Worksheet 5-13
tra ns
n -
Management Network IP Address Allocation Example 5-14
Client Access Network Configuration Worksheet 5-15 a no
a s
h eฺ
Client Access Network IP Address Allocation Example 5-16
)
b r
ฺ Guid
Optional Network Configuration Worksheets 5-17
m
k o ฺco dent
Power Distribution Unit Configuration Worksheet 5-18
t ei Stu
Auto Service Request Configuration Worksheet 5-19
o r te@ this
Oracle Enterprise Manager Grid Control Configuration Worksheets 5-20
ฺ s up use
Cell Alert Delivery Configuration Worksheet 5-21
( t ec se to
Beyond the Configuration Worksheets 5-22
e i ko licen
Configurator Spreadsheet Overview 5-23
o T
Generating the Configuration Files 5-25
ic
ecn
Database Machine Hardware Installation Overview 5-26
T Configuring Oracle Exadata Database Machine Overview 5-27
Selecting the Database Server Operating System 5-28
Deploying Solaris on the Database Servers 5-29
Reclaiming Unused Operating System Disk Space Using Linux 5-30
Reclaiming Unused Operating System Disk Space Using Solaris 5-32
Performing Initial Network Configuration 5-33
Loading the Configuration Information and Installing the Software 5-35
Running OneCommand on Database Machine 5-36
Exadata Storage Configuration 5-37
The Result After Installation and Configuration 5-39
Supported Additional Configuration Activities 5-40
Unsupported Configuration Activities 5-41
Quiz 5-43
Summary 5-45
v
6 Exadata Storage Server Configuration
Objectives 6-2
Exadata Storage Server Administration Overview 6-3
Exadata Storage Server Administrative User Accounts 6-4
dcli Overview 6-5
dcli Examples 6-6

Testing Storage Server Performance Using CALIBRATE 6-7
CALIBRATE Example 6-8
Configuring the Exadata Cell Server Software 6-9
Starting and Stopping Exadata Cell Server Software 6-10
Configuring Cell Disks 6-11
Configuring Grid Disks 6-12 le
Interleaved Grid Disks 6-13 fe rab
Interleaved Grid Disks and ASM Intelligent Data Placement 6-14
tra ns
n -
Creating Flash-Based Grid Disks 6-15
Creating Smart Flash Log 6-16 a no
a s
h eฺ
Configuring Hosts to Access Exadata Cells 6-17
)
b r
ฺ Guid
Configuring ASM and Database Instances to Access Exadata Cells 6-18
m
k o ฺco dent
Configuring ASM Disk Groups using Exadata Storage 6-19
t ei Stu
Reconfiguring Exadata Storage 6-20
o r te@ this
Optional Configuration Tasks 6-22
ฺ s up use
Exadata Storage Security Overview 6-23
( t ec se to
Exadata Storage Security Implementation 6-24
e i ko licen
Quiz 6-26
o T
Summary 6-29
ic
ecn
T Practice 6 Overview: Configuring Exadata 6-31
7 I/O Resource Management

Objectives 7-2
I/O Resource Management Overview 7-3
I/O Resource Management Concepts 7-5
I/O Resource Management Plans 7-6
I/O Resource Management Plans Example 7-7
IORM Architecture 7-10
Getting Started with IORM 7-11
Setting the IORM Objective 7-12
Enabling Intradatabase Resource Management 7-13
Intradatabase Plan Example 7-14
Enabling IORM for Multiple Databases 7-15
Interdatabase Plan Example 7-16
vi
Setting Database I/O Utilization Limits 7-17
Interdatabase Plans and Database Roles 7-18
Category Plan Example 7-19
Using Database I/O Metrics 7-20
IORM and Exadata Storage Server Flash Memory 7-21
Complete Example 7-22

Quiz 7-25
Summary 7-29
8 Recommendations for Optimizing Database Performance

Objectives 8-2 le
Optimizing Performance 8-3 fe rab
Flash Memory Usage 8-4
tr ans
n -
Compression Usage 8-6
Index Usage 8-8 a no
a s
h eฺ
ASM Allocation Unit Size 8-9 )
Minimum Extent Size 8-10 b r
ฺ Guid
m
Quiz 8-11
k o ฺco dent
Summary 8-13 t ei Stu
o r te@ this
ฺ s up use
Practice 8 Overview: Optimizing Database Performance with Exadata 8-15
( t ec se to
9 o Scan
UsingkSmart
i e n
T e
Objectives l
9-2i c
ic o Exadata Smart Scan Overview
ecn
9-3
T Smart Scan Requirements 9-4
Situations Preventing Smart Scan 9-6
Monitoring Smart Scan in SQL Execution Plans 9-7
Smart Scan Execution Plan Example 9-8
Example of a Situation Preventing Smart Scan 9-10
Smart Scan Join Processing With Bloom Filters 9-11
Smart Scan Join Filtering Example 9-12
Other Situations Affecting Smart Scan 9-13
Exadata Storage Server Statistics Overview 9-14
Exadata Storage Server Wait Events Overview 9-15
Smart Scan Statistics Example 9-16
Smart Scan Wait Events Example 9-17
Concurrent Transaction Example 9-18
Extreme Concurrent Transaction Example 9-19
Migrated Rows Example 9-20
vii
I/O Sent Directly to Database Server to Balance CPU Usage Example 9-21
Column Filtering Example 9-22
Summary 9-23
Quiz 9-24
Practice 9 Overview: Using Smart Scan 9-26
10 Consolidation Options and Recommendations

Objectives 10-2
Consolidation Overview 10-3
Different Consolidation Types 10-4
Core Principles for Database Consolidation 10-5
Recommended Consolidation Approach 10-6 le
Recommended Storage Configuration for Consolidation 10-7 fe rab
Alternative Storage Configurations 10-8
tr a ns
n -
Cluster Configuration Options 10-10 a no
Benefits and Limitations of Partitioned Storage Configurations 10-9
a s
h eฺ
Operating System Parameter Recommendations 10-11
)
Database Memory Recommendations 10-13 b r
ฺ Guid
m
CPU Management Recommendations 10-14
k o ฺco dent
Other Recommendations 10-16 t ei Stu
o r te@ this
Isolating Management Roles 10-18
ฺ s up use
Schema Consolidation Recommendations 10-20
( t ec se to
Maintenance Considerations 10-21
e i ko licen
Quiz 10-22
o T
Summary 10-24
ic
Tecn 11 Migrating Databases to Exadata Database Machine
Objectives 11-2
Migration Best Practices Overview 11-3
Performing Capacity Planning 11-4
Database Machine Migration Considerations 11-5
Choosing the Right Migration Path 11-6
Logical Migration Approaches 11-7
Physical Migration Approaches 11-9
Reducing Downtime for Migration using Transportable Tablespaces 11-11
Other Approaches 11-12
Post-Migration Best Practices 11-13
Quiz 11-14
Summary 11-16
viii
Practice 11 Overview: Migrating to Databases Machine using Transportable
Tablespaces 11-19
12 Bulk Data Loading using Oracle DBFS

Objectives 12-2
Bulk Data Loading using Oracle DBFS Overview 12-3

Preparing the Data Files 12-4
Staging the Data Files 12-5
Configuring the Staging Area 12-6
Configuring the Target Database 12-10
Loading the Target Database 12-11
Quiz 12-13 le
Summary 12-15 fe rab
tra ns
n -
Practice 12 Overview: Bulk Data Loading using Oracle DBFS 12-17
no
a
a s
h eฺ
13 Exadata Database Machine Platform Monitoring Introduction )
Objectives 13-2 b r
ฺ Guid
m
Monitoring Technologies and Standards o13-3
k ฺco dent
Simple Network Management Protocol t ei (SNMP)
S tu 13-4
Intelligent Platform Management
r @
te Interface s
hi (IPMI) 13-5
o
p u(ILOM) e t
Integrated Lights Out
ฺ s uManager s 13-6
Exadata Storage
( t ec Server e t o
Metrics, Thresholds and Alerts 13-7
Automatic
i ko Diagnostic
e s
n Repository (ADR) 13-8
T e l i c
i c o Enterprise Manager Grid Control 13-9
n Enterprise Manager Database Control 13-10
Tec Quiz 13-11
Summary 13-12
14 Configuring Enterprise Manager Grid Control 11g to Monitor Exadata Database

Machine
Objectives 14-2
Enterprise Manager Grid Control Architecture Overview 14-3
Grid Control Monitoring Architecture for Exadata Database Machine 14-4
Configuring Grid Control to Monitor Exadata Database Machine 14-5
Deploying the Oracle Management Agent 14-6
Configuring ASM and Oracle Database Targets 14-7
Deployment Overview for System Monitoring Plug-ins 14-8
Recommended Plug-in Deployment Strategy 14-9
Deploying the System Monitoring Plug-in for Oracle Exadata Storage Server 14-10
Deploying the System Monitoring Plug-in for Oracle Exadata ILOM 14-12
ix
Deploying the System Monitoring Plug-in for Oracle Exadata InfiniBand
Switch 14-13
Deploying the System Monitoring Plug-in for Exadata Cisco Switch 14-14
Deploying the System Monitoring Plug-in for Avocent MergePoint Unity (KVM)
Switch 14-15
Deploying the System Monitoring Plug-in for Oracle Exadata Power

Distribution Unit 14-16
Configuring User-Defined Metrics for Additional Network Monitoring 14-17
Configuring Grid Control for Exadata Database Machine: Alternative
Approach 14-19
Configuring the Plug-Ins for High Availability 14-20
Configuring an Exadata Database Machine Aggregate Service le
and Dashboard 14-21 fe rab
Quiz 14-23
tra ns
n -
Summary 14-25
no
a
a s
h eฺ
r )
ฺ Guid
b
15 Monitoring Exadata Storage Servers m
Objectives 15-2
k o ฺco dent
Lesson Overview 15-3 t ei Stu
e@ this15-4
Exadata Metrics and AlertsrtArchitecture
o
Monitoring Exadata u
ฺ s p Server
Storage
u se with Metrics 15-6
( t ec Cell
Monitoring Exadata
e to Examples 15-8
Metrics:
ko Exadata
Monitoring
i e s
nStorage Server with Alerts 15-9
T e l i c
i c o Monitoring Cell Alerts and Creating Thresholds: Examples 15-11
n Isolating Faults with Exadata Storage Server Quarantine 15-13
Tec Monitoring Exadata Storage Server with Active Requests 15-15
Monitoring Exadata Storage Server with Grid Control Overview 15-16
Monitoring Hardware Failure and Sensor State 15-18
Monitoring Exadata Storage Server Availability 15-19
Checking for Undelivered Alerts 15-20
Checking for Disk I/O Errors 15-21
Checking for Network Errors 15-22
Monitoring Filesystem Free Space 15-23
Comparing Metrics Across Multiple Storage Servers 15-24
Monitoring Metrics Within a Storage Server 15-25
Third Party Monitoring Tools 15-26
Quiz 15-27
Summary 15-29
Practice 15 Overview: Monitoring Exadata Storage Server 15-31
x
16 Monitoring Exadata Database Machine Database Servers
Objectives 16-2
Monitoring Database Servers Overview 16-3
Monitoring Hardware 16-4
Monitoring the Operating System 16-5
Monitoring Oracle Grid Infrastructure 16-6

Monitoring Oracle Database 16-7
Monitoring Oracle Management Agent 16-8
Quiz 16-9
Summary 16-10
17 Monitoring the InfiniBand Network le

Objectives 17-2 fe rab
InfiniBand Network Monitoring Overview 17-3
tr a ns
n -
Manually Monitoring the InfiniBand Switches 17-4
Monitoring the InfiniBand Switches with Grid Control 17-5 a no
a s
h eฺ
Monitoring the InfiniBand Switch Ports 17-6 )
b r
ฺ Guid
Monitoring the InfiniBand Ports on Database Machine Servers 17-7
m
k o ฺco dent
Monitoring the InfiniBand Fabric: Subnet Manager Master Location 17-8
t ei Stu
Monitoring the InfiniBand Fabric: Network Topology and Link Status 17-9
Quiz 17-10
o r te@ this
Summary 17-11
ฺ s up use
( t ec se to
e i ko licen
T
18 Monitoring
o Objectives Other Exadata Database Machine Components
n i c 18-2
Tec Monitoring the Cisco Catalyst Ethernet Switch 18-3
Monitoring the Sun Power Distribution Units 18-4
Monitoring the Avocent MergePoint Unity KVM Switch 18-5
Quiz 18-6
Summary 18-7
19 Other Useful Monitoring Tools

Objectives 19-2
Exachk Overview 19-3
Running Exachk 19-4
Exachk Output 19-5
DiagTools Overview 19-6
Using ADRCI on Exadata Storage Servers 19-7
Imageinfo Overview 19-8
xi
Imagehistory Overview 19-9
OSWatcher Overview 19-10
Quiz 19-11
Summary 19-13
20 Backup and Recovery

Objectives 20-2
Backup and Recovery Overview 20-3
Using RMAN with Database Machine 20-4
General Recommendations for RMAN 20-5
Disk Based Backup Strategy 20-7 le
Disk Based Backup Recommendations 20-8 fe rab
Tape Based Backup Strategy 20-10
tra ns
n -
Tape Based Backup Architecture 20-11
Tape Based Backup Recommendations 20-12 a no
a s
h eฺ
Connecting the Media Server Using Ethernet 20-14
)
Tape Based Backup Recommendations 20-15 b r
ฺ Guid
m
Hybrid Backup Strategy 20-16
k o ฺco dent
t ei Stu
Restore and Recovery Recommendations 20-17
o r te@ this
Backup and Recovery of Database Machine Software 20-18
Quiz 20-19
ฺ s up use
Summary 20-21
( t ec se to
e i ko licen
o T
Practice 20 Overview: Using RMAN Optimizations for Database Machine 20-23
ic
Tecn 21 Exadata Database Machine Maintenance Tasks
Objectives 21-2
Database Machine Maintenance Overview 21-3
Powering Database Machine Off and On 21-4
Safely Shutting Down a Single Exadata Storage Server 21-5
Replacing a Damaged Physical Disk 21-6
Replacing a Damaged Flash Card 21-8
Moving All Disks from One Cell to Another 21-9
Using the Exadata Cell Software Rescue Procedure 21-10
Quiz 21-12
Summary 21-15
22 Patching Exadata Database Machine

Objectives 22-2
Patching and Updating Overview 22-3
xii
Maintaining Exadata Storage Server Software 22-4
Maintaining Database Server Software 22-6
Assisted Patching Using OPlan 22-7
Maintaining Other Software 22-8
Recommended Patching Process 22-9
Test System Recommendations 22-11

Quiz 22-12
Summary 22-13
23 Exadata Database Machine Automated Support Ecosystem

Objectives 23-2 le
Auto Service Request Overview 23-3 fe rab
The ASR Process 23-4
tra ns
n -
ASR Requirements 23-5
Configuring ASR Manager 23-6 a no
a s
h eฺ
Configuring Exadata Database Machine for ASR 23-7
)
Activating ASR Assets 23-8 b r
ฺ Guid
m
Approve and Verify ASR Assets 23-9
k o ฺco dent
t ei Stu
Oracle Configuration Manager Overview 23-10
o r te@ this
Configuring Oracle Configuration Manager 23-11
Quiz 23-12
ฺ s up use
Summary 23-15
( t ec se to
e i ko licen
T
n i c24o Quality of Service Management
Tec Lesson Objectives 24-2
QoS Management Background 24-3
QoS Management Overview 24-4
QoS Management and Exadata Database Machine 24-5
QoS Management Focus 24-6
QoS Management Benefits 24-7
QoS Management Functional Overview 24-9
QoS Management Policy Sets 24-11
Server Pools 24-12
Performance Classes 24-14
Classification and Tagging 24-16
Performance Policies 24-17
Performance Class Ranks 24-18
Performance Objectives 24-19
Performance Satisfaction Metrics 24-20
xiii
Server Pool Directive Overrides 24-21
Overview of Metrics 24-22
QoS Management Architecture 24-24
QoS Management Recommendations 24-25
Implementing Recommendations 24-27
Quiz 24-29
Summary 24-31
A Managing Exadata Database Machine with Enterpise Manager Cloud

Control 12c
Objectives A-2 le
Lesson Overview A-3 fe rab
a
Configuring Exadata Database Machine as an Enterprise Manager Target A-4
tr ns
n -
Exadata Storage Server Monitoring A-6 a no
Visualizing Exadata Database Machine in Enterprise Manager A-5
a s
h eฺ
Exadata Storage Server Administration A-7 )
b r
ฺ Guid
Exadata Storage Server Performance Monitoring A-8
m
k o ฺco dent
Exadata Storage Server Health Monitoring A-9
t ei Stu
Exadata Database Machine InfiniBand Network Monitoring A-10
o r te@ this
Exadata Database Machine InfiniBand Network Administration A-11
Summary A-12
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn
xiv
Bulk Data Loading using Oracle DBFS
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s se and/or its affiliates. All rights reserved.
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Objectives
After completing this lesson, you should be able to:

• Describe the process of using Oracle DBFS for bulk data

loading into Database Machine
• Describe the process to configure the Database File
System (DBFS) feature for staging input data files
• Use external tables based on input data files stored in
le
DBFS to perform high-performance data loads
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Exadata Database Machine Administration Workshop 12 - 2

Bulk Data Loading using Oracle DBFS Overview
Source System Database Machine
Target Database
Staging Database CREATE ... AS SELECT ...

INSERT ... SELECT ...
DBFS le
fe rab
tr a ns
Network File
n -
no
Transfer
(FTP, NFS, etc.) Data file staged in
Database File System
s a
(DBFS)
)
External table
a
h eฺ
ฺ Guid
b r
based on
Data File m DBFS staged
k o ฺco dent data

Target tables
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T provides information on how to load databases on Database Machine with data
This lesson
o
i c
n other systems using data files as the intermediate data store. The lesson focuses on the
from
T e c
recommended approach for loading file-based data into Oracle using DBFS. This approach is
illustrated on the slide. The remainder of this lesson describes this process in greater detail.
Note that DBFS can also be as a generalized shared file system on Database Machine. You
can use the procedure for configuring DBFS contained later in this lesson to configure DBFS
for purposes other than staging data files for bulk data loading.

Preparing the Data Files
• Data files can be in any format supported by the external

table feature
• To facilitate high-performance parallel access:
– Oracle automatically divides the files into 10MB granules
— Exceptions: compressed files, data read from a pipe or a tape
– If granules cannot be created then:
— Each file is treated as a granule
r a ble
— The number of files determines the maximum degree of
n s fe
parallelism
n - tra
— no
You must use multiple files to manually enable parallelism
a
– General rules of thumb a
h eฺ s
b
ฺ themr ) id in size
— If using multiple files, then try to keep
m G u similar
—
k o ฺco dthen
If the file sizes vary significantly, e t them in order from
nlist
largest to smallest inte i external
the tu table definition
@ s S
o r te thi
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
The recommended approach for bulk data loading using Oracle DBFS relies on the external
i c
n feature of Oracle Database. The data files used for bulk data loading can be in any
table
T e c
format supported by external tables. The process for creating the data files is outside the
scope of this lesson and mostly depends on the facilities available in the source system.
However the data files are created, the following should be taken into account in order to
facilitate high-performance parallel access to the data files while they are being loaded:
• Where possible, when accessing large data files through external tables Oracle
automatically divides the files into 10 MB granules. These granules can be processed in
separate parallel processing threads. Oracle is unable to use this approach with
compressed files or data read from a pipe or a tape device.
• If granules cannot be used then each separate data file can be treated as a granule and
the number of files determines the maximum degree of parallelism that is available. You
can manually divide a large file into separate smaller files and use them to manually
enable parallelism.
• If you are using multiple input data files in conjunction with a single external table, then
you should try to keep the data files similar in size. If the file sizes do vary significantly,
then list them in order from largest to smallest in the external table definition.

Staging the Data Files
• Data files should always be staged in DBFS

– Not database server internal drives

• DBFS enables the database to be used as a file system
– Shared storage for staging or storing data files, scripts,
reports and other application files
– Files are stored as SecureFiles LOBs inside database tables
that are stored in Exadata
r a ble
– Files are exposed to the database servers using a POSIX-
n s fe
compatible file system interface
n - tra
– Files are protected like any Oracle data – ASM
o
nmirroring,
a
Data Guard, Flashback, and so on has ฺ r) uide
ฺ b
ฺ c om nt G
i k o ude
t e S t
r @
te thi s
o se and/or its affiliates. All rights reserved.
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
o T
It is recommended to stage data files inside Database Machine using Database File System
i c
n DBFS is an Oracle Database feature that enables the database to be used as a high-
(DBFS).
T e c
performance POSIX-compatible file system. Using the available space on internal database
server disk drives for staging data files is highly discouraged.
Inside DBFS files are stored as SecureFiles LOBs. A set of PL/SQL procedures implement
the file system access primitives, such as open, close, create, and so on. The dbfs_client
utility enables the mounting of a DBFS file system as a mount point on the Database Machine
database servers. It also provides the mapping from file system operations to database
operations.
Note: The procedure detailed in this lesson applies to configuring DBFS on Linux database
servers. ASM Cluster File System (ACFS) is not supported in conjunction with Exadata.

Configuring the Staging Area
DBFS should be configured in a separate staging database

• Use DBCA to create the staging database

– See My Oracle Support note 1191144.1 for
recommendations
• Create a bigfile tablespace for DBFS storage
SQL> CREATE BIGFILE TABLESPACE DBFS DATAFILE '+DBFS_DG'
SIZE 32G AUTOEXTEND ON NEXT 8G MAXSIZE 300G
a b le
NOLOGGING ONLINE PERMANENT EXTENT MANAGEMENT LOCALfe r
AUTOALLOCATE SEGMENT SPACE MANAGEMENT AUTO; rans
o n -t
• Create a DBFS user account n
a
SQL> create user dbfs identified by dbfs
) h as ฺ
quota unlimited on DBFS;
m ฺbr Guide
SQL> grant create session, create
ฺ c o table, n t create
o de
procedure, dbfs_role ito
t e k dbfs; t u
r t e @ his S
u p o se t
c ฺs © t2012,
Copyright
o uOracle and/or its affiliates. All rights reserved.
t e
i k o ( ense
While o DBFS ic
Teis fully lfunctional if it is co-located with the target database, it is recommended to
n i c
configure DBFS in a separate staging database. Use the Database Configuration Assistant
T e c
(DBCA) to create a database. If you wish to access DBFS across multiple Database Machine
database servers create the staging database as a RAC database across the desired servers.
Default settings are mostly suitable when configuring the staging database however My
Oracle Support note 1191144.1 contains the following recommendations:
• Use the General Purpose or Transaction Processing template
• Use Oracle Managed Files and the DBFS_DG disk group to store the staging database
• De-select Specify Flash Recovery Area
• Set Memory Size (SGA and PGA) to 8192 MB
• De-select Use Automatic Memory Management
• Use AL32UTF8 as the database character set
• Set the PARALLEL_MAX_SERVERS initialization parameter to 2
After the staging database is created, create a bigfile tablespace to use as the DBFS store.
The slide shows an example CREATE TABLESPACE command containing the recommended
tablespace options. Also, create a DBFS user account as suggested on the slide.

• Additional database server operating system configuration:

– Add the Oracle software owner, or user that will mount the
DBFS file system, to the fuse group
# usermod –a –G fuse oracle
– As root, create /etc/fuse.conf containing the entry:
user_allow_other
# echo "user_allow_other" > /etc/fuse.conf le
# chmod 644 /etc/fuse.conf
fe rab
tra
– Create a mount point for DBFS with ownership and group ns
n -
orouser that
permissions set to the Oracle software owner, n
a
will mount the DBFS file system as h eฺ
# mkdir /data
ฺ b r) uid
# chown oracle:dba /dataom t G
i k oฺc uden
@ te St
o r te this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T DBFS you must also perform the operating system configuration tasks listed on
To configure
o
i c
ecn
the slide. They should be performed on each database server where DBFS will be mounted.
T The examples show a typical configuration using the Oracle software owner as the DBFS
mount point owner. You can create and use an alternative operating system user if you wish
to separate DBFS access from the Oracle software owner and it’s database administration
privileges. You can also use the dcli utility to replicate the configuration steps on multiple
database servers.

• Creating the DBFS store

$ cd $ORACLE_HOME/rdbms/admin
$ sqlplus dbfs/dbfs
SQL> @dbfs_create_filesystem_advanced.sql DBFS mydbfs
nocompress nodeduplicate noencrypt non-partition
• Mounting DBFS
$ nohup $ORACLE_HOME/bin/dbfs_client dbfs@<StagingDB>
le
-o allow_other,direct_io /data < passwd.txt &
fe rab
– See My Oracle Support note 1054431.1 for automatictra ns
mounting configuration details n -
no a
• Using DBFS a
h eฺ s
b r )
ฺdirectory id
– Access DBFS through the mount m G u
– Copy files to DBFS using k o ฺco desuch
methods nt as FTP and NFS
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
After the
o Tstaging database is created and the required operating system configuration is
n i c
completed, you can create the DBFS store. Use the script located at
T e c
$ORACLE_HOME/rdbms/admin/dbfs_create_filesystem_advanced.sql. The script
must be run by the DBFS database user (created earlier in the configuration process). The
script accepts numerous parameters. In the example on the slide, DBFS represents the name
of the tablespace you created to house the DBFS store, and mydbfs represents the name of
the DBFS store. This name of the DBFS store is used later after DBFS is mounted to name
the directory that appears under the DBFS mount directory.
To mount DBFS you can use dbfs_client as shown in the example on the slide. Note that
the example references /data as the DBFS mount directory. Also, note that the DBFS user
password is contained in a file called passwd.txt. The password file is only required when
mounting DBFS and is not required while DBFS is being used. Alternatively you can configure
an Oracle Wallet so that the DBFS client can mount a DBFS store without a password.
My Oracle Support bulletin 1054431.1 contains a procedure that can be used to automatically
mount DBFS using Oracle Clusterware.

After DBFS is mounted you can access it through the mount directory (/data in the slide
example). In the mount directory you will find a subdirectory named after the DBFS store; the
contents of this subdirectory (/data/mydbfs in the example) is the contents of the DBFS
store.
Note that it is possible to have multiple copies of dbfs_client accessing the same shared
file system. The sharing and caching semantics are similar to NFS. Like NFS, the default
mode caches writes on the client and flushes them after a timeout or when the file being
modified is closed. Also like NFS, writes to a file are only visible to clients that open the file
after the writer closed the file. This behavior is commonly referred to as close-to-open cache
consistency.
For more information regarding DBFS on Database Machine refer to My Oracle Support
bulletin 1054431.1. See also the chapters on DBFS located in the Oracle Database
SecureFiles and Large Objects Developer's Guide 11g Release 2 (11.2).
le
fe rab
tra ns
n -
a no
a s
h eฺ
r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

Configuring the Target Database
• Prerequisites for data file access using external tables

– Create an Oracle directory object that references the DBFS

staging area directory
– Grant the required permissions on the Oracle directory object
– Create the required external tables
• Ensure efficient space management
– Use bigfile tablespaces le
– Use 8MB initial extents for large segments fe rab
Set the INITIAL storage parameter to 8MB tra ns
—
n -
—
a no
Set the CELL_PARTITION_LARGE_EXTENTS initialization
parameter to TRUE or ALWAYS as ) h eฺ
• Use unlimited quotas to bypass m b r
ฺ management
quota u id
o nt G
• Use the parallel clause toikset oฺctheuddefault e degree of
parallelism for the target
e
t tables S t
r t e @ h i s
u p o se t
c ฺs © t2012,
Copyright
t e
i k o ( ense
After the
o
e
TDBFS-based licstaging area is configured it can be populated with the required data
e c nicTo access the data files using external tables you must:
files.
T • Create an Oracle directory object that references the DBFS staging area directory.
• Grant the required permissions on the Oracle directory object so that the data files can
be referenced by the required target database users.
• Create the required external tables referencing the data files stored in DBFS.
You should also prepare your target database more generally to optimize the load process.
You should ensure efficient space management by following the general space management
recommendations for use in conjunction with Exadata that are listed on the slide. In addition,
you can optimize the performance of your loads if you bypass tablespace quota management
and you set an appropriate default degree of parallelism for your target tables.

Loading the Target Database
• Recommended approach uses external tables

– Parallel direct path loading for high-performance

– In-flight processing using SQL
— Transformations using SQL functions
— Sort data while loading to optimize Exadata storage indexes
– No need to re-stage the data
– Other advanced features le
For example: Input file preprocessing fe rab
—
tra ns
• SQL*Loader can also be used n -
a
– Parallel direct path loading can also be achieved
no
a s
h external
– Less processing flexibility compared r )
with d e ฺ tables
b
ฺ Gu i
– Existing SQL*Loader scriptsccanombe easily
t reused
i k oฺ uden
@ te St
o r te this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
The recommended method for loading the data files uses external tables. Loading data using
i c
n tables provides the following key benefits:
external
T e c
• You can use direct path loading and parallel processing to achieve high-performance.
• You can use the flexibility and power of SQL to efficiently process the data while it is
being loaded. For example, you can use standard or user-defined SQL functions to
transform the data during loading and you can sort the data while you load it to optimize
the storage indexes that are automatically maintained by Exadata Storage Server.
• Because you can process the data while it is being loaded, it is unlikely that you will
need to re-stage the data inside the target database.
• External tables provide some useful advanced features. A primary example is the ability
to preprocess a data file using a user-defined program which provides additional
flexibility and power to process the data using routines outside the database.
SQL*Loader can also be used to load data files. It can deliver comparable performance but
does not provide the same level of processing flexibility when compared with external tables.
SQL*Loader remains a good choice when customers already have SQL*Loader based scripts
that they wish to reuse in a Database Machine environment.

Loading the Target Database
• Parallel loading using external tables

– A CREATE TABLE ... AS SELECT statement

automatically uses the defined degree of parallelism
– An INSERT ... SELECT statement needs parallel DML to
be enabled
SQL> alter session enable parallel dml;
• Direct path loading using external tables le

fe rab
– A CREATE TABLE ... AS SELECT statement
tra ns
automatically uses direct path loading n -
– An INSERT ... SELECT statement needs a noAPPEND
an
s
hint to enable direct path loading r) ha eฺ
m ฺ my_table
u b id
c o
SQL> insert /*+ APPEND */ into
t G
koฺ den
select * from my_external_table;
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Bulk data
o Tloading using external tables is achieved using a CREATE TABLE ... AS
n i c statement or an INSERT ... SELECT statement.
T e c
SELECT
To utilize parallel processing, it is recommended that you set the default degree of parallelism
for the external tables used in the load and the target table being created or inserted into.
Alternatively, you can set parallelism within the statement using a PARALLEL hint.
After parallelism is configured, a CREATE TABLE ... AS SELECT statement will
automatically use the defined degree of parallelism. An INSERT ... SELECT statement
requires parallel DML to be enabled before it can execute in parallel.
To utilize direct path lading from external tables, you must use an APPEND hint for an
INSERT ... SELECT statement. CREATE TABLE ... AS SELECT statements
automatically use direct path loading.
Remember that the bulk data loading approach outlined in this lesson can be used in
conjunction with other techniques such as partition exchange loading.

Quiz
To facilitate parallel loading from typical flat files external table

definitions must reference numerous smaller data files rather

than one file containing all the data:
a. True
b. False
le
fe rab
tr a ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o bT
e c n
Typically, Oracle automatically divides typical flat files into 10MB granules to facilitate parallel
T processing. Oracle is unable to use this approach with compressed files or data read from a
pipe or a tape device.

Quiz
While DBFS is fully functional if it is co-located with the target

database, it is recommended to configure DBFS in a separate

staging database:
a. True
b. False
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o aT
e c n
T

Summary
In this lesson, you should have learned how to:

• Describe the process of using Oracle DBFS for bulk data

loading into Database Machine
• Describe the process to configure the Database File
System (DBFS) feature for staging input data files
• Use external tables based on input data files stored in
le
DBFS to perform high-performance data loads
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Additional Resources
• Lesson Demonstrations
– Bulk Data Loading with Database Machine

• My Oracle Support Notes
– Configuring a Database for DBFS on Oracle Database
Machine
– Configuring DBFS on Oracle Database Machine
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Practice 12 Overview:
Bulk Data Loading using Oracle DBFS
In this practice you will perform a bulk data load on Database
Machine. You will configure a database file system (DBFS) and

use it to stage a CSV formatted file. You will then use the
external table feature of Oracle Database to reference the CSV
file. Finally, you will use a CREATE TABLE AS SELECT
statement to copy the CSV file data into a table in your
database. le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Tecn
ic
oT e i
( t
ฺ s
ko licen
o r
ec se to
t
up use
k o
te@ this
ei Stu
m b r )
ฺco dent
a s
ฺ Guid
a
h eฺ
no n - tr
a
ns fe rab
le
Exadata Database Machine Platform

Monitoring Introduction
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Objectives
After completing this lesson, you should be able to describe the

key monitoring infrastructure technologies associated with

Exadata Database Machine.
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Monitoring Technologies and Standards
Exadata SNMP
Storage
Server
InfiniBand
Network
Metrics
ILOM le
Automatic
fe rab
Diagnostic
Repository tra ns
n -
no a
Enterprise
a s
h eGrid Manager
r ) d ฺ
Alerts
m b
ฺ Gu Control i
o
ฺc dent
k o
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Embeddedo T within the various Exadata Database Machine components are a variety of
n i c
monitoring technologies. Some are based on widely adopted standards while others are
T e c
specific to Database Machine. This lesson introduces the key monitoring technologies and
standards used in conjunction with Database Machine. Most of the information presented in
this lesson should be familiar to most students and it is presented to provide a consistent
foundation for the following lessons.

Simple Network Management Protocol (SNMP)
• What is it?
– A standard protocol for managing devices on a network

• What does it do?
– Primarily used to propagate information about a device to a
monitoring console somewhere on the network
— Report alerts for hardware or software issues
– Can be used to set device configuration parameters le
fe rab
• Where is it found?
t ra ns
– Throughout Database Machine, including database o n -servers,
a
storages servers, InfiniBand switches, Ethernet
nswitch, KVM
s
haPDUs
(Keyboard Video and Mouse) switchr)and
d e ฺ (Power
Distribution Units) b
mฺ Gu i
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T Management Protocol (SNMP) is an Internet-standard protocol for managing
SimpleoNetwork
i c
n on IP networks. A wide array of devices support SNMP including many servers,
devices
T e c
network switches, routers, printers and workstations. SNMP is used mostly in network
management systems to monitor network-attached devices for conditions that warrant
administrative attention.
Typically, SNMP is used to propagate information about a device to a monitoring or
management console located somewhere in the network. To do this, managed devices run an
SNMP agent. SNMP agents expose management information about the managed systems as
variables. The protocol also permits active management tasks, such as modifying and
applying a new configuration through remote modification of these variables. SNMP agents
can be separate processes or embedded into other software or hardware modules.
SNMP defines a variety of protocol data units (messages) that can be exchanged between
agents and managers. One of the most important messages is an asynchronous notification
from agent to manager known as a trap. An SNMP trap is often used by devices to report alert
conditions.
SNMP is used extensively throughout Exadata Database Machine so that the various
components of Database Machine can report monitoring information and alerts to network
management systems such as Enterprise Manager Grid Control.

Intelligent Platform Management Interface (IPMI)
• What is it?
– An open, industry-standard interface for server management

– Primarily used to perform server configuration and
management independently of the server operating system
• Where is it found?
– Exadata Database Machine database servers and Exadata rabl
e
e
Storage Servers nsf tra
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
TPlatform Management Interface (IPMI) is an open, industry-standard interface for
Intelligent
o
i c
n management of server systems over a number of different types of networks.
the
T e c
On face value it would seem that SNMP and IPMI perform essentially the same function
however there are two main differences:
• IPMI is more focused on server management. IPMI functionality includes field-
replaceable unit (FRU) inventory reporting, logging of system events and system
recovery (including system resets, power on and power off).
• IPMI is associated with an architecture that allows administrators to remotely manage a
system in the absence of an operating system or other system management software.
The monitored system may be powered off, but the baseboard management controller
(BMC) must be connected to a power source and the monitoring medium, typically a
local area network connection. The BMC is a specialized microcontroller embedded in
the server.
IPMI prescribes only the structure and format of the interfaces as a standard, while detailed
implementations may vary.
Inside Database Machine, IPMI support is built into Integrated Lights Out Manager (ILOM) on
each database server and Exadata Storage Server.

Integrated Lights Out Manager (ILOM)
• What is it?
– Integrated service processor hardware and software

– Provides out-of-band server monitoring and management to:
— Remotely control the power state of a server
— View the status of sensors and indicators on the system
— Provide a remote server console
able
– Generates alerts for hardware errors and faults as they n fe
occur
r
a s
• Where is it found? n -t r
o
nand Exadata
– Exadata Database Machine database servers a
Storage Servers ) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Integrated
o TLights Out Manager (ILOM) provides advanced service processor hardware
n i c
(baseboard management controller) and software that you can use to manage and monitor
T e c
Database Machine servers. ILOM’s dedicated hardware and software is preinstalled on the
database servers and Exadata Storage Servers inside Database Machine. ILOM enables you
to actively manage and monitor the server independently of the operating system state (out-
of-band). With ILOM, you can:
• Learn about hardware errors and faults as they occur
• Remotely control the power state of your server
• View and use the graphical and non-graphical consoles for the host
• View the current status of sensors and indicators on the system
• Determine the hardware configuration of your system
• Receive alerts about important system events through notification methods such as
SNMP traps and email alerts.
ILOM automatically initializes as soon as power is applied to the server. It provides a full-
featured, browser-based web interface and has an equivalent command-line interface (CLI).
There is also an industry-standard SNMP interface and IPMI support (ILOM is compliant with
IPMI v1.5 and v2.0).

Exadata Storage Server
Metrics, Thresholds and Alerts
• What are they?
– Metrics, thresholds and alerts provide the foundation for

monitoring Exadata Storage Server
• What do they do?
– Metrics provide a measure relating to some aspect of
storage server status or performance
– Thresholds are metric levels, which if crossed automatically
r a ble
generate an alert notification
n s fe
n - tra
– Alerts are automatically generated notifications of system
o
events an
• Where are they found? ) h as ฺ
– Exadata Storage Server m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Exadata o T
Storage Server contains a system of metrics, thresholds and alerts which provide a
n i c
foundation for storage server monitoring.
T e c
Metrics provide a measure relating to some aspect of storage server status or performance.
Exadata Storage Server includes more than 150 metrics, many of which are associated with
numerous separate measurements. For example, the current temperature of the server is a
single-value metric. In contrast, the cumulative number of requests to read large blocks from a
grid disk has a separate measurement for each grid disk. So on a typical Exadata cell this one
metric can be associated with more than 20 separate observations.
Thresholds are definitions which allow administrators to define metric levels, which if crossed
automatically generate an alert notification. Thresholds can contain two alert levels, warning
and critical. For example, a threshold could be defined to generate a warning alert when cell
memory utilization reaches 80%, and a critical alert when it reaches 90%.
Exadata Storage Server generates alerts for various system conditions. For example, an alert
is generated if the cell server software terminates unexpectedly or if a sensor detects a
potential imminent disk failure. Additional alerts can be generated for user-defined events
through the use of thresholds.
Note that Enterprise Manager also contains a separate system of metrics, thresholds and
alerts. Other software, such as Oracle Database, contains its own metrics and alerts.

Automatic Diagnostic Repository (ADR)
• What is it?
– A file-based repository for diagnostic data such as traces,

dumps and logs
– Provides a consistent organization for diagnostic data which
enables administrators and Oracle Support to correlate and
analyze diagnostic data more easily and effectively le
• Where is it found? fe rab
tr a ns
-
– Exadata Database Machine database servers andnExadata
o
Storage Servers an
) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T Diagnostic Repository (ADR) is a file-based repository for diagnostic data such
The Automatic
o
i c
asntraces, dumps, the alert log, health monitor reports, and more. It has a unified directory
e c
T structure across multiple products and multiple instances. Beginning with Release 11g, the
database, Automatic Storage Management (ASM), listener, and other Oracle products store
diagnostic data in the ADR. Exadata Storage Server also uses the ADR structure to organize
its diagnostic data.
Each instance of each product stores diagnostic data under its own home directory within the
ADR. For example, in an Oracle Real Application Clusters environment with shared storage
and Oracle ASM, each database instance and each ASM instance have separate ADR home
directories.
ADR's unified directory structure, consistent diagnostic data formats across products and
instances, and a unified set of tools enable customers and Oracle Support to efficiently
correlate and analyze diagnostic data across multiple instances.

Enterprise Manager Grid Control
Grid Control
Console
Agents
and
plug-ins
Oracle
Management
Service le
fe rab
tra ns
n -
no a
Oracle a
h eฺ s
Management
Repository ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Enterprise
o TManager Grid Control is a system management software platform that delivers
n i c
centralized monitoring, administration, and life cycle management functionality for information
T e c
technology infrastructure, including systems running Oracle and non-Oracle technologies.
When used in conjunction with Exadata Database Machine, Enterprise Manager Grid Control
should be configured with a set of system-specific plug-ins to enable monitoring of the various
Database Machine components. Additionally, there are best practice recommendations
regarding how to configure Database Machine targets inside Grid Control in order to facilitate
monitoring of Database Machine as a unified whole rather than as a collection of discrete
components. The next lesson in this course describes how to configure Grid Control for
monitoring Exadata Database Machine.

Enterprise Manager Database Control
Database Control
Console
Database
Control
Processes
and
Repository
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
TManager Database Control is a database administration environment built into
Enterprise
o
i c
n Database. Database Control, can perform administrative tasks such as creating
Oracle
T e c
schema objects (tables, views, indexes, and so on), managing user security, managing
database memory and storage, performing database backup and recovery, and bulk data
movement (export and import). Performance and status information about the database can
also be viewed.
In addition to its core database administration functions, Database Control also provides
monitoring and management facilities for Exadata Storage Servers. However, Database
Control does not provide the breadth of monitoring capabilities offered by Grid Control. For
this reason, Grid Control is the recommended monitoring environment for Exadata Database
Machine. In addition, Database Control is not considered further during this course.

Quiz
Which of the following monitoring infrastructure technologies

can be used in conjunction with Exadata Storage Servers:

a. Simple Network Management Protocol (SNMP)
b. Integrated Lights Out Manager (ILOM)
c. Exadata metrics, thresholds and alerts
d. Automatic Diagnostic Repository (ADR)
le
e. Enterprise Manager Grid Control fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o a,T b, c, d, e
e c n
T

Summary
In this lesson, you should have learned how to describe the key
monitoring infrastructure technologies associated with Exadata

Database Machine.
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Configuring Enterprise Manager Grid Control

11g to Monitor Exadata Database Machine
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Objectives

• Describe the Enterprise Manager Grid Control architecture

as it specifically applies to Exadata Database Machine.
• Describe the placement of agents, plug-ins and targets
• Describe the recommended configuration for high
availability
• Describe the plug-ins associated with Exadata Database erab
le
Machine and how they are configured. n s f
n - tra
• Describe how to configure a Dashboard for Exadata
a no
Database Machine. has r ) e ฺ
b
ฺ Gu i d
o m
k o ฺc dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Enterprise Manager Grid Control
Architecture Overview
Grid Control
Console
Agents
and
plug-ins
Oracle
Management
Service le
fe rab
tr a ns
n -
no a
Oracle a
h eฺ s
Management
Repository ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
i c o TManager Grid Control contains four major elements:
Enterprise
e c n1. Oracle Management Service (OMS) is the processing heart of the system.
T 2. Oracle Management Repository is the persistent store of monitoring, management and
configuration information.
3. Grid Control Console provides a web-based management interface.
4. Agents and plug-ins are deployed to managed targets to collect information which is
processed by the OMS.
This fundamental architecture does not change when Grid Control is used to monitor Exadata
Database Machine.

Grid Control Monitoring Architecture
for Exadata Database Machine
Oracle Exadata Database Machine Enterprise Manager
Grid Control 11g
Release 1
Exadata Storage
Servers Grid Control
Cisco
Sun InfiniBand Console
Ethernet
Switch(es) CELLSRV MS
Switch
SNMP CellCLI
traps over SSH
Oracle Database Servers
Oracle
le
DB Instance Management XML
rab
Agent 11g over
Release 1 HTTP/
Avocent Oracle
ASM Instance
HTTPS
fe
ns
MergePoint Management
InfiniBand Exadata
Unity KVM
Listener
Plug-in Plug-in
Service
- tr a
Switch
no n
Clusterware Cisco
Plug-in
PDU
Plug-in
s a
Sun
) a
h eฺ
Power
Host Information
ILOM ฺ Guid
b r
KVM Oracle
m
ฺco dent
Distribution Plug-in Plug-in Management
ILOM Interface Repository
Units
k o
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Wheno T
Enterprise Manager Grid Control is used to monitor Exadata Database Machine, an
i c
n Management Agent should be deployed to every database server. The agent is used
Oracle
e c
T monitor and maintain Oracle software targets on the database server including database
to
instances, ASM instances, listeners, clusterware resources and other host information.
In addition, a series of plug-ins are available which extend Grid Control to enable monitoring
of other Database Machine components such as the Exadata Storage Servers, InfiniBand
switches, Ethernet switch, KVM switch and Power Distribution Units (PDUs).
Each plug-in connects to its monitoring targets using different methods. For example, the
Exadata Storage Server plug-in gathers monitoring data from the cells by making CellCLI
calls over SSH, however Exadata cell alerts are delivered directly to the agent using SNMP
traps. Each plug-in is described in greater detail later in this lesson.
Note that some of the plug-ins only support Enterprise Manager Grid Control 11g Release 1
(or later) and Oracle Management Agent 11g Release 1 (or later). Hence these are the
recommended minimum versions to use in conjunction with Database Machine.

Configuring Grid Control to Monitor
Exadata Database Machine
1. Deploy Oracle Management Agent to all database servers
2. Configure ASM and Oracle Database targets

3. Deploy the system monitoring plug-ins for Database Machine
– Exadata Storage Server
– Integrated Lights Out Manager (ILOM)
– InfiniBand Switch(es)
– Cisco Switch le
– Avocent MergePoint Unity (KVM) Switch fe rab
– Power Distribution Units ns tra
n -
4. Configure user-defined metrics for additional network
a no monitoring
5. Configure the plug-ins for high availability has ฺ
6. Configure an Exadata Database Machine ฺ b r) aggregate
u ide service and
m
co ent G
dashboard o ฺ
t eik Stud
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The list o inTthe slide outlines the recommended steps to configure Enterprise Manager Grid
i c
n for monitoring Exadata Database Machine. The assumed starting point is after the
Control
T e c
initial installation and configuration of both Database Machine and Grid Control.
The remainder of this lesson describes each step in greater detail.
Recorded demonstrations showing an example of each process step are also provided in
conjunction with this lesson. A full list of the recorded demonstrations is provided in the
additional resources section at the end of the lesson.

Deploying the Oracle Management Agent
• Where to deploy:
– All Database Machine database servers

• Recommended version:
– 11g Release 1 (11.1) or later
• How to deploy:
– Grid Control Agent Installation Wizard
– See demonstration Agent Installation and Configuration r a ble
n s fe
– Manual installation of some OS packages may be required
ra
o n -t
a n
) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
It is recommended to install Oracle Management Agent 11g Release 1 (or later) on all the
n i c
Database Machine database servers.
T e c
Assuming that Enterprise Manager Grid Control 11g Release 1 is already running and there is
network connectivity between it and the Database Machine, the easiest way to deploy the
agent is to use the Grid Control agent installation wizard. An example of this process is
provided in the demonstration entitled Agent Installation and Configuration.
Some earlier versions of the Database Machine database servers were shipped without all
the packages required by the Oracle Management Agent. It is acceptable to modify the
database servers to install the missing packages.
In the event that the Oracle Management Agent fails to install as a result of missing packages,
install the required packages and retry the agent installation. A list of packages required for
the agent is provided in the Enterprise Manager documentation.

Configuring ASM and Oracle Database Targets
• Management targets are automatically discovered after the

Oracle Management Agent is started

• ASM instances and Oracle databases are discovered but
are listed as unconfigured
• Administrators must supply monitoring credentials to
complete configuration
• The procedures are essentially the same regardless of r a ble
whether or not Exadata Database Machine is involvednsfe
• See demonstration Configuring ASM and Database n - tra
a no
Targets as
r ) h eฺ
m ฺ Guid
b
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
After the
o TOracle Management Agent is installed and running on each database server,
n i c
management targets are automatically discovered. This includes ASM instances, cluster
T e c
databases, database instances, clusters, clusterware resources, network listeners, host
information and of course the agents themselves.
After the initial target discovery phase, the ASM and Oracle Database instances are
discovered but are listed as unconfigured. To complete the configuration of the ASM and
Oracle Database instances, administrators must supply monitoring credentials.
For each ASM instance, connectivity between Grid Control and the ASM instance must be
configured using the credentials of an administrative user having either the sysdba or
sysasm role.
For each database, the dbsnmp user account must be enabled and its password must be
supplied to enable connectivity from Grid Control. Note that to enable the dbsnmp account
requires sysdba access to the database.
Note that the procedure for configuring the ASM instance targets is the same regardless of
whether or not the instances reside on Exadata Database Machine. Likewise, the same
procedure is used to configure database instance targets on Database Machine as on other
clustered database environments.
An example of how to configure the ASM and database targets is provided in the
demonstration entitled Configuring ASM and Database Targets.
Deployment Overview for
System Monitoring Plug-ins
• Plug-ins extend Grid Control to enable monitoring of
additional hardware and software components

• Each plug-in contains:
– Logic that defines how Grid Control communicates with the
component and what metrics are monitored
– User Interface definitions that extend the Grid Control
Console le
• Plug-in implementation process: fe rab
tr ans
1. Perform prerequisites n -
2. a no
Import the plug-in into the Grid Control environment
) h as ฺ
3.
ฺbr Guide
Deploy the plug-in to one or more agents
m
4. Add a target for each instance
o ฺ co of ethe
n t component being
monitored eik t ud S t
r @
te thi s
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
System o T
monitoring plug-ins extend Enterprise Manager Grid Control, enabling it to monitor
n i c
additional hardware and software component types. For Exadata Database Machine, a pack
T e c
of suitable plug-ins is available. Each plug-in contains logic which enables a management
agent to communicate with the type of system component being monitored. It also contains
user interface elements which extend the Grid Control Console and provide administrators
with a view of the information associated with the plug-in.
The implementation process for each plug-in contains the following steps:
1. Perform prerequisites. Because plug-ins communicate with different system components
in different ways, they each have different prerequisites which must be configured. The
prerequisites are documented in the Installation Guide which accompanies each plug-in.
2. Import the plug-in. Plug-ins are packaged as Java archives (jar files). Before a plug-in
can be used, its package must be imported into the Grid Control environment.
3. Deploy the plug-in. Deployment configures the plug-in on one or more agents. This
enables the agent to communicate with the type of hardware or software component
associated with the plug-in.
4. Add targets. After a plug-in is deployed, each component instance must be registered as
a target.
Details and additional information for each plug-in follows later in this lesson.

Recommended Plug-in Deployment Strategy
• Recommendation:
– Associate all plug-in targets with a primary agent on one of

the Database Machine database servers
– Configure target failover using a secondary agent on another
database server for high availability
• Rationale:
– At any time, all plug-in targets are associated with one agent ble
which simplifies management and problem diagnosis s fera
t r a n
– Automatic target failover ensures continuation of monitoring
if the primary agent is down n on-
– The performance impact of monitoring h
s a confined to
isalow and
r ) d e ฺ
one database server m b
ฺ Gu i
– Storage Server monitoring can
o
ฺc leverage
n t the InfiniBand
k o e
interconnect t ei Stud
@ r te this
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
Subjecto Tversion requirements and network connectivity, the various Exadata monitoring
to
i c
n can be deployed to any management agents associated with the Grid Control
plug-ins
T e c
environment. For example, it is possible (though not recommended) for the Exadata Storage
Server plug-in to be deployed against an agent running on the OMS server.
The best practice recommendation for Database Machine plug-in deployments is to deploy all
the plug-ins to at least two agents running on different Database Machine database servers.
Administrators are advised to configure all the plug-in targets so that they are all initially
associated with just one of the management agents, the primary agent. After the initial round
of targets is configured, there is a recommended process to configure the plug-ins for high
availability. The resulting configuration automatically migrates all the plug-in targets to a
nominated secondary agent in cases where the primary agent goes down. The slide lists the
rationale behind this recommendation. Configuring the plug-ins for high availability is
described later in this lesson.
The plug-in configuration examples and descriptions which follow throughout this lesson are
all based on the above recommendation. Customers that choose an alternative plug-in
deployment strategy will need to adjust their configurations accordingly.

Deploying the System Monitoring Plug-in for
Oracle Exadata Storage Server
• Prerequisites:
– Preferred credentials for the agent targets

– SSH user equivalence between the agent user and the
cellmonitor user on all the cells
• Recommendations:
– Configure a realm to enable aggregated reporting
le
CellCLI> alter cell realmName = <Realm Name>
fe rab
ra ns
– Configure the cells to deliver alerts to Grid Control
o n -t
using
SNMP a n
s ha eฺ
• Demonstration: r
ฺb Guid )
m
– Configuring the Exadata Storage
k o ฺco dServer e nt Plug-in
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T prerequisites are required to successfully deploy the System Monitoring Plug-in
The following
o
i c
fornOracle Exadata Storage Server:
T e c
• Preferred credentials must be set in the Grid Control preferences. The host username
and password for the user running the Management Agent must be specified. Preferred
credentials must be set for each agent host where the plug-in is deployed.
Note that Agent Preferred Credentials are also required to successfully deploy other
system monitoring plug-ins. Typically, Preferred Credential settings are made once prior
to deploying the first plug-in and are reused for subsequent plug-ins.
• The storage server plug-in communicates with Exadata cells using the cellmonitor
user account on the Exadata Storage Servers. SSH user equivalence must be
configured between the user running the Management Agent on the agent host (typically
oracle) and the cellmonitor user on all the cells that will be monitored by the plug-
in. Because the cellmonitor user account has limited access to the storage server
operating system, a special procedure is required to establish SSH user equivalence.
This procedure is documented in the Installation Guide that accompanies the storage
server plug-in. An example is also shown in the demonstration that accompanies this
topic.

In addition, the following recommendations apply:
• Exadata Storage Servers can be collected into a group known as a realm. The Exadata
Storage Server plug-in contains numerous reports that aggregate monitoring information
for a realm. Typically all of the cells in one Exadata Database Machine will be grouped
into a realm however this is not an absolute requirement. It is recommended that you
define realms based on logical groupings that make sense in your environment.
• In conjunction with the plug-in, it is recommended that you configure all your cells to
deliver alerts to Grid Control using SNMP. This involves enabling the SNMP notification
method on each cell along with adding the agent host name and port as an SNMP
subscriber.
An example of how to deploy the storage server plug-in is provided in the demonstration
entitled Configuring the Exadata Storage Server Plug-in.
le
fe rab
tra ns
n -
a no
a s
h eฺ
r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

Oracle Exadata ILOM
• Recommendation:
– Use the ILOM plug-in only for monitoring ILOM on Database

Machine database servers
— Not required for Exadata Storage Server because ILOM
monitoring is already managed by Exadata Storage Server
software.
• Prerequisites:
le
– ipmitool version 1.8.10.1 or later
fe rab
– ILOM user credentials ns tra
o n -
— A dedicated ILOM user account for the plug-in isnhighly
a
recommended
) h as ฺ
• Demonstration:
m ฺbr Guide
– Configuring the ILOM Plug-in
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The ILOM
o Tplug-in monitors the ILOM service processor in a server. It enables hardware
i c
n and sensor data to be monitored using Grid Control. For Database Machine, it is
events
T e c
recommended to use the ILOM plug-in only for monitoring the ILOM server processor of the
Database Machine database servers. The ILOM plug-in is not required to monitor Exadata
Storage Servers since ILOM metrics and alerts are already managed by the Exadata Storage
Server software.
The ILOM plug-in requires ipmitool software version 1.8.10.1 or later to be present on the
servers being monitored by the plug-in.
Though not an absolute requirement, it is highly recommended to create a specific ILOM user
account for the ILOM plug-in to use. The procedure is documented in the Installation Guide
that accompanies the ILOM plug-in.
An example of how to deploy the ILOM plug-in is provided in the demonstration entitled
Configuring the ILOM Plug-in.

Oracle Exadata InfiniBand Switch
• Prerequisites:
– Switch firmware version 1.1.3-2 or later

– SNMP configuration:
— Enable SNMP
— Set SNMP protocol to v1 (version 1)
— Set the SNMP community string if you plan to use an SNMP
community string other than the default (public)
le
• Demonstration: fe rab
– Configuring the InfiniBand Switch Plug-in tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The following
o
i c
fornOracle Exadata InfiniBand Switch:
T e c
• The InfiniBand Switch must be running firmware version 1.1.3-2 or later.
• SNMP must be configured on the InfiniBand Switch to allow the plug-in to obtain its
information. Here is a summary of the required settings:
- SNMP must be enabled
- The SNMP protocol must be set to v1 (version 1)
- If you plan to use an SNMP community string other than the default (public),
then you must make the appropriate setting and make sure you use the same
setting when you later create the InfiniBand switch targets.
An example of how to deploy the InfiniBand switch plug-in is provided in the demonstration
entitled Configuring the InfiniBand Switch Plug-in.

Exadata Cisco Switch
• Prerequisites:
– Switch configuration:
— Enable access to allow the plug-in to poll the switch
— Configure the switch to deliver SNMP traps to the plug-in host
— Configure the switch to send only environmental monitor SNMP
traps
Verify and save the settings
—
a b le
– An SNMP trap forwarder on each plug-in host to catch Cisco fe r
switch traps and forward them to the Management Agent n s
n - tra
• Demonstration:
a no
– Configuring the Cisco Ethernet Switch h
) as ฺ
Plug-in
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The following
o
i c
fornExadata Cisco Switch:
T e c
• The Cisco switch must be configured to enable SNMP communications between the
switch and the plug-in. The slide provides and overview of the required configuration
steps. Note that the Cisco Ethernet switch provided with Database Machine does not
include SSH support. Oracle recommends connecting to the switch management
interface using an alternative such as telnet.
• The Cisco switch has no documented way to send SNMP traps to ports other than UDP
162. However the Oracle Management Agent runs as the oracle user and cannot
listen on UDP ports under 1024. A trap forwarder must be configured so that the SNMP
traps generated by the switch are forwarded to the Oracle Management Agent.
Detailed instructions for both prerequisite tasks are provided in the Installation Guide that
accompanies the plug-in. An example of how to deploy the Cisco Ethernet switch plug-in is
provided in the demonstration entitled Configuring the Cisco Ethernet Switch Plug-in.

Avocent MergePoint Unity (KVM) Switch
• Prerequisites:
– Switch software version 1.2.8 or later

– Switch configuration:
— Confirm that SNMP is enabled on the KVM switch
— Configure the switch to deliver SNMP traps to the plug-in host
– An SNMP trap forwarder on each plug-in host to catch the
KVM traps and forward them to the Management Agent
able
Same configuration as required for the Cisco switch plug-in fer
—
a n s
• Demonstration: n -t r
– Configuring the Avocent KVM Switch Plug-ina n
o
) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
i c o T prerequisites are required to successfully deploy the KVM switch plug-in:
The following
e c n• The KVM plug-in requires that the application software version on the KVM switch is
T 1.2.8 or later. Confirm you have the required software version.
• Confirm that SNMP is enabled on the KVM switch.
• Configure the KVM switch to send SNMP traps to the host where the KVM switch plug-in
will be deployed.
• The KVM switch sends SNMP traps to the default port (UDP 162). However the Oracle
Management Agent runs as the oracle user and cannot listen on UDP ports under
1024. Configure a trap forwarder to send SNMP traps generated by the KVM on to the
Oracle Management Agent. If you have already configured a trap forwarder for the Cisco
switch plug-in, no additional configuration is required.
Detailed instructions for all prerequisite tasks are provided in the Installation Guide that
accompanies the plug-in. An example of how to deploy the KVM switch plug-in is provided in
the demonstration entitled Configuring the Avocent KVM Switch Plug-in. An example showing
the required SNMP trap forwarder configuration is contained in the Cisco switch plug-in
deployment demonstration.

Oracle Exadata Power Distribution Unit
• Prerequisites:
– PDU connected and configured for management network access

– PDU firmware version 1.02 or later
– SNMP enabled on PDU
– PDU configured to communicate with the plug-in host
– PDU alarm thresholds set. Settings based on:
Database Machine Version:
le
—
— Exadata Database Machine X2-2 (using X4170 and X4275 servers)

fe rab
—
tr a ns
Exadata Database Machine X2-2 (using X4170 M2 and X4270 M2 servers)
Exadata Database Machine X2-8 (using X4800 and X4270 M2 servers)
—
n -
— Database Machine Size: no a
— Full Rack, Half Rack or Quarter Rack
a
h eฺ s
PDU Type:
—
r )
ฺ Gu15idkVA Three Phase
b
Low-voltage 15 kVA Single Phase m Ӎ Low-voltage
ฺco Ӎ dHigh-voltage
—
High-voltage 15 kVA Single Phase

—
k o e nt 15 kVA Three Phase
t ei PhaseStuӍ High-voltage 24 kVA Three Phase
Low-voltage 24 kVA Three
—
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The following
o
i c
fornOracle Exadata Power Distribution Unit (PDU):
e c
T • By default, the PDUs inside Exadata Database Machine are not physically connected to
any network. To monitor PDUs using the plug-in, connect each PDU to the management
network and configure the PDU network port with IP addressing information. Note that
each PDU should be connected directly to your network, not to the Cisco Ethernet
switch in the Database Machine rack. This configuration helps to ensure that alerts
generated by the PDUs are delivered to Grid Control even if a power problem causes a
fault in the Database Machine management network.
• Confirm that the PDU firmware version is 1.02 or later. The current version can be
obtained by logging into the web interface of the PDU. On the left side of the screen,
click Module Info to view the PDU firmware version.
• Confirm that SNMP is enabled on the PDU and configure the PDU to enable SNMP
communications with the host where the PDU plug-in will be deployed.
• Configure the PDU with alarm threshold values. Appropriate threshold values depend on
the Oracle Exadata Database Machine version, the Oracle Exadata Database Machine
size, and the PDU type used. The Oracle Exadata Database Machine Owner's Guide
contains the recommended threshold values for each situation.
Refer to the PDU User's Guide and the Installation Guide that accompanies the PDU plug-in
for further details.
Configuring User-Defined Metrics
for Additional Network Monitoring
• The health of InfiniBand network ports should be
monitored on all Database Machine servers

– InfiniBand monitoring is automatically performed by MS on
storage servers
• User Defined Metric (UDM) scripts are available from My
Oracle Support note 1110675.1:
– emudm_ibconnect.sh monitors InfiniBand connectivity to
r a ble
other Database Machine servers
n s fe
– emudm_netif_state.sh monitors network interface
n - trastate
o
an
for Ethernet and InfiniBand network interfaces
s
•
r ha eMetrics
See demonstration Configuring User )Defined ฺ for
Additional Network Monitoring mฺ b i d
Gu ฺco dent
k o
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
InfiniBand
o Tport monitoring checks the health of InfiniBand network ports and interfaces on
n i c
database servers and Exadata Storage Servers. InfiniBand port monitoring on storage servers
T e c
is performed by the Exadata Storage Server Management Server (MS). No additional
monitoring is required on storage servers. If a port is not functioning correctly, MS creates an
alert, and delivers the alert via SNMP to Grid Control.
By default, Database Machine database servers have no built-in InfiniBand monitoring.
Checks using commands such as ibstatus and perfquery can be run periodically,
however this is a manual process. These checks can be automated using User Defined
Metrics (UDMs) in Grid Control. There are two UDM scripts provided for database servers:
• emudm_ibconnect.sh - This UDM script monitors connectivity over the InfiniBand
network to other database servers and storage servers. This list of database servers is
built from the ocrdump SYSTEM.crs.e2eport key. The list of cells is built from
cellip.ora. The script does not validate connectivity to additional devices on the
InfiniBand network such as media servers.
Note that the ibhosts command is not used because it requires root permissions.
The approach used in this script is preferred as scope is limited to servers that are
relevant to the Exadata Database Machine, and not the whole InfiniBand network.
Using this script it is recommended to create one UDM for each Database Machine
database server.

• emudm_netif_state.sh - This UDM script can monitor the network interface state for
Ethernet or InfiniBand interfaces. For InfiniBand interfaces, it checks the underlying
InfiniBand port, state, rate, and 3 error counters; RcvErrors, SymbolErrors, and
LinkIntegrityErrors. If the interface is the master for Linux bonding, then the state
of the slave interfaces is verified. The error counters are reported only when they
increase since the last invocation of the script and only when LinkDowned has not
increased. Previous values are cached in the temporary file
/var/tmp/ibCounterCache_<device><port>. Oracle recommends using this

script to create one UDM for each network interface on each Database Machine
database server.
The UDM scripts are available from My Oracle Support note 1110675.1. An example of how
to deploy the UDM scripts is provided in the demonstration entitled Configuring User Defined
Metrics for Additional Network Monitoring.
le
fe rab
tra ns
n -
a no
a s
h eฺ
r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

Configuring Grid Control for Exadata Database
Machine: Alternative Approach
• There is an alternative method for rapidly installing
Grid Control and the Exadata plug-ins. It can:

– Install Grid Control on a standalone server
– Install the management agent and Exadata plug-ins on:
— A newly established Grid Control installation
— An existing Grid Control environment with other existing targets
– Remove the management agent from database servers le
– Remove Grid Control from a standalone server fe rab
tr a ns
• The approach is integrated with the Database Machine n -
installation and initial configuration processes a no
) h as ฺ
• Linux and Solaris database server b OS
r uide are both
options
ฺ
supported com t G
oฺ uden
i
te Stk
r @
te this
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
T procedure exists for rapidly installing Grid Control and the Exadata plug-ins.
An alternative
o
i c
ecn
The procedure is integrated with the Database Machine installation and initial configuration
T process and supports both the Linux and Solaris database server OS options. The procedure
is outlined in the updated Oracle Exadata Database Machine Owner's Guide and requires
patch 11852882.
The procedure provides the ability to install a fresh Grid Control environment on to a
standalone server using a one-step installation program called setupem.sh. In addition,
setupem.sh provides functionality to install the Exadata plug-ins on a fresh Grid Control
installation or on an existing Grid Control environment which may already be monitoring other
targets. This includes performing the prerequisites for each plug-in, importing the plug-in into
Grid Control and deploying the plug-in to all of the Database Machine management agents.
Removal of the Exadata plug-ins and the Grid Control software can also be performed.
Note that while setupem.sh can deploy the Exadata plug-ins, administrators must still
configure all the Grid Control targets which represent the various Database Machine
components.

Configuring the Plug-Ins for High Availability
• Normally a plug-in target is bound to a specific agent

– If the agent is down the target cannot be monitored

• A procedure exists to facilitate target failover to a secondary
agent
– See My Oracle Support note 1110675.1
– See demonstration Configuring Plug-ins for High Availability
Oracle Exadata Database Machine
le
Database Server A Database Server B
fe rab
Primary Secondary
tr a ns
Management
Agent
Management
Agent
n -
To To
no a
Targets
Plug-in
Targets
Plug-in
h a s Oracle
Plug-in Plug-in
) ฺ
ฺbr Guide
Plug-in Plug-in Management
Plug-in Plug-in
Service
o m t
i k oฺc uden
@ te St
o r te this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T a target monitored using a plug-in is bound to the agent used during initial
By default,
o
i c
ecn
deployment. A procedure exists which allows the monitoring of a plug-in target to be
T automatically moved to a predefined secondary agent in cases where the primary agent is
down. This is done by configuring each target to have a default host and a failover host.
When the agent on the default host, the primary management agent, becomes unavailable for
any reason, monitoring of the affected targets is automatically moved to the agent on the
failover host, the secondary management agent. When the primary management agent
becomes available again, the targets are automatically moved back to it.
To enable target failover and failback you must install the failover extension PL/SQL package
into the management repository. The failover configuration process is described in the
document entitled Installation of plug-ins and dashboard configuration. This document and the
failover extension are available from My Oracle Support note 1110675.1.
Note that when plug-ins are configured for high availability, the prerequisites for each plug-in
must be met for both the primary agent host and the secondary agent host. For example, you
must define Agent Preferred Credentials for both the primary and secondary management
agent hosts. Also, you must deploy the plug-ins to both hosts.
An example of how to configure the plug-ins for high availability is provided in the
demonstration entitled Configuring Plug-ins for High Availability.

Configuring an Exadata Database Machine
Aggregate Service and Dashboard
Targets Systems Services Aggregate Dashboard
Service
ASM
ASM ASM ASM
Instances
Instances System Service
See demonstration
Hosts Hosts Hosts Creating a Dashboard for Database Machine
Hosts System Service
Storage
Storage Storage Storage Database Database
Servers
Servers System Service Machine Machine
Service Dashboard
Listeners Listener Listener
Listeners System Service
le
Database
Database
Instances
Instances
Database
System
Database
Service
fe rab
InfiniBand
InfiniBand
tra ns
Switches
Switches
InfiniBand
System
InfiniBand
Service
n -
no
a
PDUs
PDUs PDU System PDU Service
a
h eฺ s
ฺ Guid
b r )
Cisco Switch Cisco System Cisco Service
m
k o ฺco dent
KVM KVM System
ei Stu
KVM Service
t
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
After all
o T Exadata system monitoring plug-ins are installed and configured, Grid Control can
the
i c
benused to monitor all the Exadata Database Machine components (targets). For a Full-Rack
e c
T Database Machine there are typically more than 60 targets.
Naturally, administrators prefer to monitor Database Machine as a single entity rather than 60
separate components. A recommended approach has been developed to facilitate monitoring
of Database Machine as a single entity. It is based on best-practice fundamentals however
there is scope in some areas for customization to suit different needs. The diagram in the
slide illustrates the resulting hierarchy of targets inside Grid Control. Following is an outline of
the high-level methodology involved:
1. For each Database Machine target type in Grid Control, create a collection containing all
the targets of that type using the System target type. For example, group the ASM
instances into one System, group the Exadata Storage Servers into another System,
and so on.
2. Create a Service based on each of the previously created Systems. A Service is an
abstraction of a System that can have administrator-defined performance and usage
metrics associated with it. Different performance and usage metrics can be created to
suit different situations. For example, an administrator of a Data Warehouse might be
interested in I/O throughput as a key measure of performance while an administrator of
an OLTP database might be more interested in an average response time indicator.

3. Create an Aggregated Service to group all the Services into a single Grid Control entity.
The screen shot on the slide shows an example of the home page for an Aggregate
Service in Grid Control.
4. Create a monitoring Dashboard based on the Aggregate Service. The Dashboard
provides a single screen overview of the health, performance and usage of the
Database Machine. The Dashboard can be customized to suit varying requirements by
displaying different performance and usage metrics.
The process outlined above is detailed in the document entitled Installation of plug-ins and
dashboard configuration. This document is available from My Oracle Support note 1110675.1.
An example of how to configure a dashboard for Exadata Database Machine is provided in
the demonstration entitled Creating a Dashboard for Database Machine.
le
fe rab
tra ns
n -
a no
a s
h eฺ
r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

Quiz
What is the recommended implementation approach for

Exadata Database Machine system monitoring plug-ins?

a. Configure all the plug-in targets using different agents to
distribute the monitoring load
b. Configure each plug-in target using all the available agents
for maximum availability
c. Configure all the plug-in targets using just one agent for rable
the sake of simplicity n s fe
d. Configure all the plug-in targets using one primary n - tra
agent
a no
and configure target failover to a secondary a s agent to
facilitate high availability and simplifyr ) h
deployment e ฺ
ฺ b i d u
o m t G
i k oฺc uden
@ te St
o r te this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o dT
e c n
T

Quiz
You can customize the Exadata Database Machine target in

Enterprise Manager Grid Control 11g by displaying different

performance and usage metrics:
a. TRUE
b. FALSE
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o bT
e c n is no Enterprise Manager Grid Control 11g target type for Database Machine. Rather
There
T you aggregate the component targets into Systems and Services to provide a consolidated
view of the entire Database Machine.

Summary

• Describe the Enterprise Manager Grid Control architecture

as it specifically applies to Exadata Database Machine.
• Describe the placement of agents, plug-ins and targets
• Describe the recommended configuration for high
availability
• Describe the plug-ins associated with Exadata Database erab
le
Machine and how they are configured. n s f
n - tra
• Describe how to configure a Dashboard for Exadata
a no
Database Machine. has r ) e ฺ
b
ฺ Gu i d
o m
k o ฺc dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

– Agent Installation and Configuration

– Configuring ASM and Database Targets
– Configuring the Exadata Storage Server Plug-in
– Configuring the ILOM Plug-in
– Configuring the InfiniBand Switch Plug-in
– Configuring the Cisco Ethernet Switch Plug-in
– Configuring the Avocent KVM Switch Plug-in r a ble
– n s
Configuring User Defined Metrics for Additional Network Monitoring
fe
– Configuring Plug-ins for High Availability n - tra
a no
– Creating a Dashboard for Database Machine s
• My Oracle Support Notes r ) ha eฺ
m ฺb Guid
– Oracle Database Machine Monitoring
k o ฺco deBest nt Practices
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Monitoring Exadata Storage Servers
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Objectives

• Describe Exadata Storage Server metrics, alerts and

active requests
• Identify the recommended focus areas for Exadata
Storage Server monitoring
• Describe how to monitor the recommended Exadata
le
Storage Server focus areas
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Lesson Overview
• Exadata Metric and Alerts Architecture and Overview

• Monitoring Exadata Storage Server with Grid Control

Overview
• Monitoring Hardware Failure and Sensor State
What to Monitor
• Monitoring Exadata Storage Server Availability
• Checking for Undelivered Alerts
a b le
• Checking for Disk I/O Errors fe r
n s
• Checking for Network Errors
n - tra
• Monitoring Filesystem Free Space a no
• Comparing Metrics Across Multiple Storage ) h as Servers
ฺ
b r
ฺ Gu i d e
• Monitoring Metrics Within a Storagem
c enServer
o t
k o ฺ
• ei Stud
Third Party Monitoring tTools
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T focuses on best practice recommendations for Exadata Storage Server regarding
This lesson
o
i c
n should be monitored from a general administrative perspective, and how it should be
what
T e c
monitored.
The list in the slide summarizes the topics contained in the lesson. The shaded area
highlights the recommended areas to monitor.

Exadata Metrics and Alerts Architecture
MS keeps a
set of the metric values.
Collected metrics: One hour of LIST METRICCURRENT

Cell, Cell Disks,
Grid Disk, IORM, in-memory
Interconnect
Metric metric values
thresholds
CELLSRV
exceeded
internal
errors
Every hour MS
CELLSRV
ADR flushes metric
collects CELLSRV MS values to disk.
metrics
le
fe rab
ns
Cell ALTER CELL
Disk
tra
software Seven days
issues
Email
1 hour o n -
metrics
Cell and/or LISTnMETRICHISTORY

SNMP a
Cell
hardware
) h as ฺ LIST ALERTHISTORY
alerts
issues
Metric
m ฺbrandGAlert u ide
k o ฺco dHistory e nt
t ei Stu
Admin
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T in the slide illustrates the architecture for Exadata Storage Server metrics and
The diagram
o
i c
n It shows how CELLSRV periodically records important run-time properties, called
alerts.
T e c
metrics, for cell components such as CPUs, cell disks, grid disks, flash cache, and IORM
statistics. These metrics are recorded in memory. Based on its own metric collection
schedule, the Management Server (MS) gets the set of metric data accumulated by
CELLSRV. MS keeps a subset of the metric values in memory, and writes a history to an
internal disk-based repository every hour. This process is conceptually similar to database
AWR snapshots.
The retention period for metric and alert history entries is specified by the
metricHistoryDays cell attribute. You can modify this setting with the CellCLI ALTER
CELL command. By default, it is seven days. You can view the metric value history by using
the CellCLI LIST METRICHISTORY command, and you can view the current metric values by
using the LIST METRICCURRENT command.
At the Exadata cell level, you can define thresholds for a selection of metrics prior to Exadata
Storage Server software release 11.2.2.3.1. Commencing with Exadata Storage Server
software release 11.2.2.3.1, thresholds can be defined for all cell metrics.

In addition to collecting metrics, Exadata Storage Server can generate alerts. Alerts represent
events of importance occurring within the cell, often indicating that an Exadata cell function is
compromised. MS generates an alert when it discovers a:
• Cell hardware issue
• Cell software or configuration issue
• CELLSRV internal error
• Metric that has exceeded a threshold defined in the cell

You can view previously generated alerts using the LIST ALERTHISTORY command. In
addition, you can configure the cell to automatically send an email and/or SNMP message to
a designated set of administrators.
le
fe rab
tra ns
n -
a no
a s
h eฺ
r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

Monitoring Exadata Storage Server with Metrics
Metric
alertState metricObjectName unit objectType metricValue metricType name …
normal number cumulative

warning % (percentage) instantaneous
critical F (fahrenheit) rate
le
rab
C (celsius) transition
fe
ns
CREATE|ALTER THRESHOLDS
IORM_CONSUMER_GROUP
IORM_DATABASE
- tr a
Thresholds IORM_CATEGORY
no n
CELL
s a
CELLDISK
) a
h eฺ
name
ฺ Guid
b
CELL_FILESYSTEM r
comparision
m
critical
occurances
k o ฺco dent
GRIDDISK
observation
t ei StuHOST_INTERCONNECT
te@ this
warning FLASHCACHE
o r
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Metrics o T recorded observations of important run-time properties or internal instrumentation
are
i c
n of the storage cell and its components, such as cell disks or grid disks. Metrics are a
values
T e c
series of measurements that are computed and retained in memory for a period of time, and
stored on a disk for a more permanent history.
The graphic in the slide describes some of the important metric attributes. Each metric:
• Has a name and description
• Is associated with a metricObjectName that is the name of the object being
measured, such as a specific cell disk, grid disk, or consumer group
• Belongs to a group that is defined by its objectType attribute. The possible groups are
shown in the slide.
• Has a metricType, which is an indicator of how the statistic was created or defined.
Possible values and their meanings are:
- cumulative: Cumulative statistics since the metric was created
- instantaneous: Value at the time that the metric is collected
- rate: Rates computed by averaging statistics over observation periods
- transition: Are collected at the time when the value of the metrics has
changed, and typically captures important transitions in hardware status
• Has a measurement unit. Possible units are shown in the slide.

Understanding the composition of the metric name provides a good insight into the meaning
of the metric. The value of the name attribute is a composite of abbreviations. The attribute
value starts with an abbreviation of the object type on which the metric is defined:
• CL_ (cell)
• CD_ (cell disk)
• GD_ (grid disk)
• FC_ (flash cache)

• DB_ (database)
• CG_ (consumer group)
• CT_ (category)
• N_ (interconnect network)
After the abbreviation of the object type, many metric names conclude with an abbreviation
that relates to the description of the metric. For example, CL_FANS is the instantaneous le
number of working fans on the cell.
fe rab
a ns
I/O-related metric name attributes continue with one of the following combinations to identify
tr
the operation: n -
• IO_RQ (number of requests)
no
a
a s
h eฺ
• IO_BY (number of MB)
r )
ฺ Guid
b
• IO_TM (I/O latency) m
• IO_WT (I/O wait time) k o ฺco dent
Next in the name could be _R for read t
or
e_Wi for write.
S tu Following that, there might be _SM or
r @ s
hi At the end of the name, there could be _SEC
terespectively.
_LG to identify small or large I/Os,
o e t
s
to signify per second or ฺ_RQupto signify
usper request. For example:
• CD_IO_RQ_R_SM
c
(te isnthe t o
enumber of requests to read small blocks on a cell disk.
o s
eik lice is the number of MB of large block I/O per second on a grid disk.
• GD_IO_BY_W_LG_SEC
T
co value crosses a user-defined threshold, an alert will be generated. Metrics can be
If a imetric
n
Tecassociated with warning and critical thresholds. Thresholds relate to extreme values in the
metric, which might indicate a problem or other event of interest to an administrator.
Thresholds are supported on cell disk and grid disk I/O error count metrics
(CD_IO_ERRS_MIN and GD_IO_ERRS_MIN), along with the cell memory utilization
(CL_MEMUT ) and cell filesystem utilization (CL_FSUT) metrics. In addition, you can set
thresholds for I/O Resource Management (IORM) related metrics. The CellCLI LIST
ALERTDEFINITION command lists the metrics for which thresholds can be set.
For further details regarding Exadata cell metric and threshold attributes, refer to the Oracle
Exadata Storage Server Software User's Guide.

Monitoring Exadata Cell Metrics: Examples
CellCLI> LIST METRICDEFINITION WHERE objectType ='CELL' DETAIL

name: CL_CPUT
description: "Cell CPU Utilization is the percentage of time over

the previous minute that the system CPUs were not
idle (from /proc/stat). "
metricType: Instantaneous objectType: CELL
unit: %
...
CellCLI> LIST METRICHISTORY WHERE name like 'CL_.*' –

AND collectionTime > '2009-10-11T15:28:36-07:00'
le
rab
CL_RUNQ cell03_2 6.0 2009-10-11T15:28:37-07:00
CL_CPUT cell03_2 47.6 % 2009-10-11T15:29:36-07:00
fe
CL_FANS cell03_2 1 2009-10-11T15:29:36-07:00
tra ns
CL_TEMP cell03_2 0.0 C 2009-10-11T15:29:36-07:00
n -
CL_RUNQ
...
cell03_2 5.2 2009-10-11T15:29:37-07:00
no a
a
h eฺ s
CellCLI> LIST METRICCURRENT WHERE objectType = 'CELLDISK'
ฺ Guid
b r )
CD_IO_TM_W_SM_RQ CD_1_cell03 205.5 us/request
m
CD_IO_TM_W_SM_RQ CD_2_cell03 93.3
ฺco us/request
CD_IO_TM_W_SM_RQ CD_3_cell03ko 0.0 us/request e n t
... t ei Stud
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The slide
i c o Tshows some basic commands that you could use to display metric information:
e c n• Use the LIST METRICDEFINITION command to display the metric definitions for the
T cell. A metric definition describes the configuration of a metric. The example does not
specify any particular metric, so all metrics corresponding to the WHERE clause are
printed. In addition to the WHERE clause, you can also specify the metric definition
attributes you want to print. If the ATTRIBUTES clause is not used, a default set of
attributes is displayed. To list all the attributes, you can add the DETAIL keyword at the
end of the command.
• Use the LIST METRICHISTORY command to display the metric history for the cell. A
metric history describes a collection of past metric observations. Similar to the LIST
METRICDEFINITION command, you can specify attribute filters, an attribute list, and
the DETAIL keyword for the LIST METRICHISTORY command. The above example
lists metrics having names that start with CL_ that were collected after the specified
time.
• Use the LIST METRICCURRENT command to display the current metric values for the
cell. The above example lists all cell disk metrics. The metric values shown in the slide
correspond to the average latency per request of writing small blocks to a cell disk. For
this metric there is a metric observation for every cell disk.

Monitoring Exadata Storage Server with Alerts
Alert
alertSource severity alertType metricObjectName examinedBy metricName
name
BMC warning stateful
ADR critical stateless alertAction
Metric clear
info alertMessage
able
fe r
n s
failedMail
a
ALTER ALERTHISTORY ALL examinedBy="<administrator>" nfailedSNMP -t r
n o
a
) h as ฺ beginTime
m ฺbr0 Guide EndTime
o t
oฺc uden2
1
t i
e Stk notificationState
r t e @ his 3 …
u p o se t
c ฺs © t2012,
Copyright
t e
i k o ( ense
Exadatao T e lic alerts represent events of importance occurring within the storage
Storage Server
e c nictypically indicating that storage cell functionality is either compromised or in danger of

cell,
T failure. An administrator should investigate alerts, because they might require urgent
corrective or preventative action. Notification of cell alerts using email or SNMP, can be
configured during the initial configuration of Database Machine. An existing configuration can
be adjusted, or a new configuration can be established, at any time using the ALTER CELL
CellCLI command.
Cell alerts can be stateful or stateless. Stateful alerts represent observable cell states that can be
subsequently retested to detect whether the state has changed, indicating that a previously
observed alert condition is no longer a problem. Stateless alerts represent point-in-time events
that do not represent a persistent condition; they simply show that something occurred.
Alerts can have one of the following severities: warning, critical, clear, or info. Examples
of possible events that trigger alerts are physical disk failure, disk read/write errors, cell
temperature exceeding recommended value, cell software failure, and excessive I/O latency.
Metrics can be used to signal stateful alerts using warning or critical threshold values.
When the metric value crosses the threshold value, an alert is signaled. An alert with a clear
severity indicates that a previous critical or warning condition has returned to normal. For
threshold-based alerts, a clear alert is generated when the measured value crosses back
over the threshold value.

Alerts with an info severity are stateless and log conditions that might be informative to an
administrator but for which no administrator action is required. Informational alerts are not
distributed by email or SNMP notifications.
The slide illustrates some of the important alert attributes. Each alert has the following
attributes:
• name provides an identifier for the alert.
• alertSource provides the source of the alert. Some possible sources are listed in the
slide.
• severity determines the importance of the alert. Possible values are warning,
critical, clear, and info.
• alertType provides the type of the alert: stateful or stateless. Stateful alerts are
automatically cleared on transition to normal. Stateless alerts are never cleared unless
you change the alert by setting the examinedBy attribute. This attribute identifies the
administrator who reviewed the alert and is the only alert attribute that can be modified le
by the administrator using the ALTER ALERTHISTORY command.
fe rab
a ns
• metricObjectName is the object for which a metric threshold has caused an alert.
tr
n -
no
• metricName provides the metric name if the alert is based on a metric.
a
a s
• alertAction is the recommended action to perform for this alert.
h eฺ
r )
• alertMessage provides a brief explanation of the alert.
ฺ Guid
b
m
• failedMail is the intended email recipient when a notification failed.
k o ฺco dent
ei Stu
• failedSNMP is the intended SNMP subscriber when a notification failed.
t
o r te@ this
• beginTime provides the timestamp when an alert changes its state.
ฺ s up use
• endTime provides the timestamp for the end of the period when an alert changes its
state.
( t ec se to
• notificationState indicates progress in notifying subscribers to alert messages:
e i ko licen
o T
- 0: never tried
ic
ecn
- 1: sent successfully
T - 2: retrying (up to 5 times)
- 3: five failed retries
Note: Some I/O errors may result in an ASM disk going offline without generating a cell alert.
You should continue to perform I/O monitoring from your databases and ASM environments
to identify and remedy these kinds of problems.

Monitoring Cell Alerts and
Creating Thresholds: Examples
CellCLI> LIST ALERTDEFINITION ATTRIBUTES name, metricName, description

ADRAlert "CELL Incident Error"
HardwareAlert "Hardware Alert"
StatefulAlert_CG_IO_RQ_LG CG_IO_RQ_LG "Threshold Based Stateful Alert"
StatefulAlert_CG_IO_RQ_LG_SEC CG_IO_RQ_LG_SEC "Threshold Based …Alert"
StatefulAlert_CG_IO_RQ_SM CG_IO_RQ_SM "Threshold Based Stateful Alert"
...
le
rab
CellCLI> LIST ALERTHISTORY WHERE severity = 'critical' -
AND examinedBy = '' DETAIL
fe
tra ns
n -
CellCLI> ALTER ALERTHISTORY 1671443814 examinedBy="JFV" no a
a
h- eฺs
CellCLI> CREATE THRESHOLD ct_io_wt_lg_rq.interactive r
ฺ Gu- id
b )
o nt
warning=1000, critical=2000, comparison='>', m
occurrences=2, observation=5 ฺc
k o e
t ei Stud
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The slide
o Tshows some examples of CellCLI commands that display alert information. The
n i c
commands for displaying alerts are very similar to the ones used for displaying metric
T e c
information:
• Use the LIST ALERTDEFINITION command to display the definition for every alert
that can be produced on the cell. The example in the slide displays the alert name,
metric name, and description. The metric name identifies the metric on which the alert is
based. ADRAlert and HardwareAlert are not based on any metric and, therefore, do
not have metric names.
• Use the LIST ALERTHISTORY command to display the alert history for a cell. The
example in the slide lists in detail all critical alerts that have not been reviewed by an
administrator.
• Use the ALTER ALERTHISTORY command to update the alert history for the cell. The
above example shows how to set the examinedBy attribute to the user ID of the
administrator that examined the alert. The examinedBy attribute is the only
ALERTHISTORY attribute that can be modified. The example uses the alert sequence ID
to identify the alert. alertSequenceID provides a unique sequence ID number for the
alert. When an alert changes its state, another occurrence of the alert is created with the
same sequence number but with a different timestamp.

• The CREATE THRESHOLD command creates a threshold that specifies the conditions for
generation of a metric alert. The example creates a threshold for the CT_IO_WT_LG_RQ
metric associated with the INTERACTIVE category. This metric specifies the average
number of milliseconds that large I/O requests issued by the category have waited to be
scheduled by IORM in the past minute. A large value indicates that the I/O workload
from this category is exceeding the allocation specified for it in the IORM plan. The alert
is triggered by two consecutive measurements (occurrences=2) over the threshold
values: one second for a warning alert (warning=1000) and two seconds for a critical
alert (critical=2000). The observation attribute is the number of measurements
over which measured values are averaged.
le
fe rab
tra ns
n -
a no
a s
h eฺ
r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

Isolating Faults with
Exadata Storage Server Quarantine
Quarantine
ioGridDisk ioBytes ioOffset quarantineType quarantineReason name
dbUniqueID cellsrvChecksum
dbUniqueName SQL PLAN clientPID

DISK REGION
ble
planLineID Database crashReason
Cell Offload
SQLID fe r a
remoteHostName
n
creationTime s
n - tra
rpmVersion
n o incidentID
a
comment
ALTER QUARANTINE 109 -
) h as ฺ sqlID
comment = 'Relates to Bug XYZ'
e
ฺbr GuidsqlPlanHashValue
m
LIST QUARANTINE WHERE
k o ฺco- dent
t ei StuDETAIL
sqlid = '5xnjp4cutc1s8'
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Tto metrics and alerts, when prescribed faults are detected within Exadata Storage
In addition
o
i c
n a quarantine object is automatically created. By means of this, the action that caused
Server,
e c
T the fault can be quarantined so that the fault can be avoided in the future. Quarantine
of
reduces the chance of storage server software crashes, and improves storage availability.
For example, if the cell crashes while performing Smart Scan for a SQL statement, Exadata
Storage Server quarantines the SQL statement. Later, when the same SQL statement occurs
again, the cell will not allow the SQL statement to use Smart Scan.
The following types of automatic quarantine are available:
• SQL PLAN: Created when the cell crashes while performing Smart Scan for a SQL
statement. The SQL Plan for the SQL statement is quarantined, and Smart Scan is
disabled for the SQL plan.
• DISK REGION: Created when the cell crashes while performing Smart Scan of a disk
region. The 1 MB disk region being scanned is quarantined and Smart Scan is disabled
for the disk region.
• Database: Created when the cell detects that a particular database causes instability.
Instability detection is based on the number of SQL Plan Quarantines for a database.
Smart Scan is disabled for the database.

• Cell Offload: Created when the cell detects some offload feature has caused
instability. Instability detection is based on the number of Database Quarantines for a cell.
Smart Scan is disabled for all databases.
When a quarantine is created, an alert is generated to notify administrators about what was
quarantined, why the quarantine was created, when and how the quarantine can be dropped
manually, and when the quarantine is dropped automatically. All quarantines are automatically
removed when a cell is patched or upgraded.
The Exadata Storage Server quarantine mechanism is designed to be automated and self-
contained. However, CellCLI commands are available to manually manipulate quarantines.
• Use the LIST QUARANTINE command to show the quarantines currently on the cell.
• Use the ALTER QUARANTINE command to set the comment attribute. The comment
attribute is the only quarantine attribute that can be modified.
• Use the DROP QUARANTINE command to manually remove a quarantine. In general, a
quarantine can be removed if the quarantined entity is not expected to cause further le
problems. Refer to the alert message associated with the quarantine for more detailsfe rab
before manually removing a quarantine.
tra ns
n -
• Use the CREATE QUARANTINE command to manually create a quarantine object. Manual
a no
creation of quarantines should be done in coordination with Oracle Support Services.
a s
h eฺ
Manual quarantines are created to proactively isolate SQL statements which are known to
r )
ฺ Guid
b
cause problems. Following is an example of manual quarantine creation:
m
k o ฺco dent
CELLCLI> CREATE QUARANTINE quarantineType="SQLID",
sqlid="5xnjp4cutc1s8"
t ei Stu
sqlPLanHashValue attribute isoonly r te@
The slide lists the attributes associated with theisquarantine object. Note that the
th when the quarantineType is SQL PLAN.
applicable
Likewise, the IO attributesฺs u p
(ioGridDisk,
e
us ioBytes and ioOffset) are only applicable when
the quarantineType
c
(teis DISK t o
e REGION.
o n s
T eik lice
n i co
Tec

Monitoring Exadata Storage Server
with Active Requests
Active Request
ioGridDisk ioBytes ioOffset ioReason ioType objectNumber id name
parentID asmDiskGroupNumber
requestState asmFileIncarnation
file initialization
sessionID
read
write
asmFileNumber
able
predicate pushing
fe r
sessionSerNumber filtered backup read consumerGroupName
a n s
predicate push read
n r
-t dbName
tablespaceNumber
n o
a
sqlID
) h as instanceNumber
ฺ
b
ฺ Gu r i d e
LIST ACTIVEREQUEST -
o m t
fileType
WHERE IoType = 'predicate
k o ฺc pushing'
d e n -
DETAIL i
te Stu
r @
te this
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
An activeTrequest provides a client-centric or application-centric view of client I/O requests
that
n i careo currently being processed by a cell.
Tec The slide shows the most important attributes of an active request. You can see that an active
request is characterized at all levels: instance, database, ASM, and cell. Most of the attributes
have self-explanatory names. Here is a brief explanation of some of the attributes:
• ioReason is the reason for the I/O activity, such as a control-file read.
• ioType identifies the type of active request. Possible values are listed in the slide.
• requestState identifies the state of the active request. Possible values include:
- Accessing Disk - Computing Result
- Network Receive - Network Send
- Queued Extent - Queued for Disk
- Queued for File Initialization - Queued for Filtered Backup Read
- Queued for Network Send - Queued for Predicate Pushing
- Queued for Read - Queued for Write
- Queued in Resource Manager
Use the LIST ACTIVEREQUEST command to display active request details for the cell. The
syntax is very similar to other LIST commands. You can specify which attributes to display or
you can display them all using the DETAIL clause. You can also filter the output using a
WHERE clause.

with Grid Control Overview
• Each storage server is a separate target in Grid Control
– Storage servers can be grouped together in a System

• Metrics inside Grid Control are mostly based on cell metrics
• Additional thresholds can be set in Grid Control
• Alerts generated inside the cell are displayed inside Grid Control
• Additional alerts can be generated inside Grid Control
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T Storage Server is monitored as a single Grid Control target, which covers the
An Exadata
o
i c
ecn
hardware, operating system and Exadata Storage Server software. Exadata Storage Servers
T are independent units that are each identified as separate targets in Grid Control. However,
storage servers can be grouped together in a System to facilitate monitoring as a group.
Grid Control in conjunction with the Exadata Storage Server plug-in leverage and extend the
system of metrics and alerts within each Exadata Storage Server.
When Grid Control is used to monitor storage servers, there are two different types of related
metrics: storage server metrics and Enterprise Manager metrics. In most cases there is a one-
to-one mapping between the two. The Management Server (MS) on Exadata Storage Server
collects, computes, and manages metrics for the individual storage server. The Exadata
Storage Server plug-in gathers metrics, typically from all the Database Machine storage
servers, and presents them to the user in Grid Control as Enterprise Manager metrics.
In addition to cell-level metric thresholds, using the Exadata Storage Server plug-in you can
set separate Grid Control thresholds for all the Exadata metrics supported by the plug-in. Prior
to Exadata Storage Server software release 11.2.2.3.1, cell thresholds can only be set on a
limited set of cell metrics so Grid Control thresholds offer the opportunity to automatically
raise alerts based on a wider range of metrics.

When equivalent metrics can be defined in either Grid Control or at the cell level, the general
recommendation is to define the threshold at the cell level so that the threshold definition and
any associated alerts are maintained within the cell.
All Exadata Storage Server alerts are delivered by the storage server to Grid Control using
Simple Network Management Protocol (SNMP). There are two types of server alerts that
come from Exadata Storage Server:
• For Integrated Lights Out Manager (ILOM)-monitored hardware components, ILOM

reports a failure or threshold exceeded condition as an SNMP trap, which is received by
MS. MS processes the trap, creates an alert for the storage server, and delivers the alert
via SNMP to Grid Control.
• For MS-monitored hardware and software components, MS processes the failure or
threshold exceeded condition, creates an alert, and delivers the alert via SNMP to Grid
Control.
From an end-user perspective there is no difference between these two kinds of alerts.
r a ble
An alert message may contain a corrective action which needs to be performed to resolve
n s fe the
alert. For example, the circled area in the screen shot on the slide indicates thera
action to take
- t
for the specified alert.
n on
The fix for bug 8814019 should be installed to resolve incorrect text
s ain the Action portion of
Exadata Storage Server alerts. ) a
h eฺ
r
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
i c oT
e c n
T

Monitoring Hardware Failure and Sensor State
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
CellCLI> list alerthistory t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T of an Exadata Storage Server is monitored collectively by ILOM and MS. ILOM
The hardware
o
n i c
monitors availability and sensor state using preset thresholds for Exadata Storage Server
T e c
hardware components such as the system motherboard, processors, memory, power
supplies, fans, and network interface controllers. MS monitors other hardware components
directly, including the disk controller, hard disk drives, flash accelerator cards, and InfiniBand
host channel adapter (HCA).
Together ILOM and MS provide full hardware monitoring and alerting. When an issue arises
an alert is automatically generated within the cell, and is also propagated to Grid Control (if
configured). The screen-shot in the slide shows an example of an alert notification for a
hardware fault.
If Grid Control or some other SNMP manager is not configured, then administrators should
configure cells to deliver alerts using email or they should periodically check the cell using the
CellCLI LIST ALERTHISTORY command.

Monitoring Exadata Storage Server Availability
le
fe rab
tra ns
n -
no a
a
h eฺ s
CellCLI> list cell detail
ฺ Guid
b r )
<Output Truncated>
m
cellsrvStatus: running
k o ฺco dent
msStatus:
t i
running
erunning tu
rsStatus:
@ s S
o r te thi
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Monitoring
o T of Exadata Storage Server availability is built into the Exadata Storage Server
i c
n for Grid Control. The screen shots in the slide show examples of how Grid Control
plug-in
T e c
indicates that storage servers are available or unavailable.
For environments without Grid Control, storage server availability should be checked by
confirming network connectivity. In addition, the status of the cell services should be checked
using the CellCLI LIST CELL DETAIL command. The output for a healthy cell should
contain the following:
cellsrvStatus: running
msStatus: running
rsStatus: running

Checking for Undelivered Alerts
• Periodically check for undelivered alerts

– CellCLI example checking one cell:

CellCLI> list alerthistory where notificationState != 1 and examinedBy = ''
1_1 2011-04-25T12:09:22-07:00 warning "The warning
threshold for the following metric has been crossed. Metric Name :
CL_MEMUT Metric Description : Percentage of total physical memory on the cell
that is currently used Object Name : exa1cel01 Current Value :
51.0 % Threshold Value : 50.0 % "
le
rab
CellCLI>
fe
– dcli example that checks all the cells listed in the
a ns
cell_group file: n-tr no
a
as ฺ
$ dcli -g cell_group cellcli -e "list alerthistory where notificationState != 1
and examinedBy = \'\'"
) h
m ฺbr Guide
• If undelivered alerts exist, check
k o ฺco cell-to-agent
e n t network
connectivity, agent availability t ei Sand t ud cell configuration
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Relyingo T Grid Control for alert notifications depends on alerts being reliably propagated to
on
i c
n Control. While this is usually the case, administrators should periodically check for
Grid
T e c
undelivered alerts.
The CellCLI LIST ALERTHISTORY command can be used to check for undelivered alerts.
The first example in the slide shows how to check for alerts that don't have a success status
(1) and which have not been marked as examined by an administrator. The second example
shows how to use the same command in conjunction with dcli to check a group of cells at
once.
Note that an alert might not have a success state for many reasons and that output generated
by the monitoring command in the slide does not necessarily relate to a cell fault or
misconfiguration. For example, an alert might not be propagated because the agent is
unavailable. Also, if the cell is configured to deliver alerts using SNMP and email, it is possible
for the alert to show an unsuccessful notification state when one delivery method succeeds
but the other one fails. Finally, if communication is disrupted between a storage server and
the agent where the Exadata Storage Server plug-in is deployed, alerts processed by MS may
not be delivered to Grid Control. Remember that MS will retry up to 5 times to deliver the alert,
so if the communication disruption is temporary you may observe an undelivered alert and
then see the same alert delivered moments later.

Checking for Disk I/O Errors
• On each cell, create a warning threshold using the

following CellCLI CREATE THRESHOLD command:

CellCLI> create threshold CD_IO_ERRS_MIN comparison='>', warning=0, -
> occurrences=1, observation=1
Threshold CD_IO_ERRS_MIN successfully created
CellCLI> list threshold CD_IO_ERRS_MIN detail

name: CD_IO_ERRS_MIN
comparison: >
observation: 1
le
occurrences:
warning:
1
0.0
fe rab
tra ns
CellCLI>
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The slide
o Tshows a recommended storage server threshold setting that assists in checking for
i c
n I/O errors. Disk I/O errors are typically handled by Exadata cell software automatically.
disk
T e c
The recommended threshold setting causes an alert to be generated by the cell when an I/O
error is detected.
No action is required for a single warning alert. However many warnings for a single drive
may by a precursor to a drive failure. Likewise, many warnings across all the drives in a single
cell may indicate an impending fault with a controller or some other component.
If this warning is reported and a critical alert is also generated by MS for a disk component,
follow the action specified in the critical alert.

Checking for Network Errors
• Monitor the storage network for dropped packets:

– Using Grid Control

— For the storage server target metrics:
— Host MB Dropped Per Sec
— Host RDMA MB Dropped Per Sec
— Apply the following settings:
Warning threshold to Zero
le
—
— Collection Schedule to Repeat Every 5 Minutes

fe rab
— Upload Interval to 3 Collections
tra ns
– Directly on the cell n -
a
Monitor the Host Interconnect Metrics, in particular:no
as ฺ
—
N_MB_DROP_SEC
) h
ฺbr Guide
—
— N_MB_RDMA_DROP_SEC
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T storage network is essential to the overall health and performance of Exadata
A functioning
o
i c
ecn
Database Machine. A loss of functionality is often indicated by the amount of dropped network
T traffic. Normally, dropped network traffic is automatically dealt with inside Database Machine
so occasional dropped packets are not a cause for concern. However many dropped packets
for a single cell may be a precursor to a problem.
To raise the awareness of such situations it is recommended to set thresholds inside Grid
Control for Host MB Dropped Per Sec and Host RDMA MB Dropped Per Sec as described on
the slide. Note that the settings must be applied to each storage server target.
Periodic monitoring of the N_MB_DROP_SEC and N_MB_RDMA_DROP_SEC cell metrics is
advised for environments without Grid Control. The use of thresholds and alerts to monitor
these metrics at the cell level can be adopted in preference to setting Grid Control thresholds
using Exadata Storage Server software release 11.2.2.3.1 and beyond.

Monitoring Filesystem Free Space
• Filesystem free space on Exadata Storage Servers is

monitored automatically by MS:

– No direct administrator action is required
– MS generates an alert if free space becomes low
– MS automatically reclaims used space by purging old log
files, trace files, crash dumps and other unnecessary files
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Exadatao T
Storage Server software automatically monitors filesystem utilization on each storage
i c
n There is no need to set a metric threshold for filesystem utilization within the cell or in
server.
T e c
Grid Control. MS monitors filesystem free space, generates critical alerts when free space
becomes low, and takes corrective action to free used space. For more detailed information
see the section entitled Understanding Automated Cell Maintenance in the Oracle Exadata
Storage Server Software User's Guide 11g Release 2 (11.2).

Comparing Metrics Across
Multiple Storage Servers
• Compare resource utilization across storage servers using cell
metric observations and realm-wide reports in Grid Control

• Large imbalances may indicate a problem
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Exactly o T data is spread across a Database Machine depends upon many factors including
how
i c
n number of applications and databases that are consolidated on the environment, along
the
T e c
with how storage is configured to meet various performance, availability, security and other
requirements.
However, in general terms work performed by storage servers to satisfy database requests
should be evenly spread across multiple storage servers. A significant imbalance in resource
utilization or I/O performance for a single storage server compared to the others should be a
cause for further investigation.
Various CPU, memory and I/O utilization metrics can be periodically compared across the
storage servers to ensure there are no unwarranted imbalances. Grid Control users can utilize
the various realm-wide reports to quickly and easily perform comparisons across a group of
storage servers. The screen shot in the slide shows and example of the realm performance
report which provides a very useful high-level comparison for key metrics across a realm of
storage servers.

Monitoring Metrics Within a Storage Server
• There are no default metric thresholds for CPU, memory or I/O

utilization
• 100% utilization may be normal in certain cases
• Metric observations are useful and relevant when compared with
previously captured baselines
• Suggested methodology:
1. Determine key metrics based on your applications and service
requirements r a ble
2. Record baseline metric observations based on normal and peak n s fe
system usage n - tra
a no outside
3. Implement thresholds to generate alerts for observations
s
expected ranges ) ha eฺ ฺbr Guid
Warning thresholds to indicate slightly abnormal observations
m
ฺco dent
—
Critical thresholds to indicate situations that could affect service levels

o
—
t k
ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Exadatao T
Storage Server is designed to be a well balanced storage subsystem. As a result,
i c
n memory, and I/O metric values on Exadata Storage Server can be higher than a typical
CPU,
T e c
database or general-purpose server, but still be in proper operating range.
For example, CPU may be 100% utilized while performing queries against data using Exadata
Hybrid Columnar Compression (EHCC). In this case, 100% CPU utilization is not problematic
and no alert should be generated.
Resource usage metric observations within a storage server are only useful and relevant if
they are compared with values captured during a period of normal operation. The slide
outlines a suggested methodology to determine important metrics and implement thresholds
based on previously recorded observations.

Third Party Monitoring Tools
• Installing additional software, including any third party

monitoring agent, is not supported on Exadata Storage

Server
• Exadata Storage Server can be configured to send alerts
to any SNMP subscriber
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
It is noto T
permissible to install any additional software, including third party monitoring agents,
i c
onnExadata Storage Server.
T e c
It is permissible to configure Exadata Storage Server to send alerts using SNMP to third party
management consoles.

Quiz
When using Grid Control to monitor Exadata Storage Servers,

where should thresholds be defined?

a. In Grid Control only
b. Inside the Exadata cell only
c. Inside the Exadata cell if possible, otherwise in Grid
Control
d. Either in Grid Control or inside the Exadata cell, it doesn't erab
le
matter n s f
ra n -t
n o
a
) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o cT
e c n equivalent metrics can be defined in either Grid Control or at the cell level, the general
When
T recommendation is to define the threshold at the cell level so that the threshold definition and
any associated alerts are maintained within the cell.

Quiz
Assuming that the Exadata Storage Server plug-in is properly

configured, which of the following can occur?

a. A metric observation generates an alert in the cell but no
alert is seen in Grid Control
b. A metric observation generates an alert in the cell and an
alert is seen in Grid Control
c. A metric observation generates no cell alert but an alert is rable
seen in Grid Control n s fe
d. A metric observation results in two alerts in Grid n - tra
Control
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o a,T b, c, d
An
c
T metric observation generates an alert in the cell but no alert is seen in Grid Control: This
e can occur when a cell-based threshold triggers a cell alert but the alert is not propagated to
Grid Control due to a network communication problem.
A metric observation generates an alert in the cell and an alert is seen in Grid Control: This is
normal and expected behaviour. It occurs when a cell-based threshold triggers an alert and
the alert is propagated to Grid Control.
A metric observation generates no cell alert but an alert is seen in Grid Control: This occurs
when a threshold set in Grid Control triggers an alert, and there is either no cell-level
threshold for the metric or the cell-level threshold is set at a level which doesn't trigger an alert
in the cell.
A metric observation results in two alerts in Grid Control: This occurs when a cell-level
threshold and a Grid Control threshold are defined on the same metric, and the corresponding
cell-level alert is propagated to Grid Control.

Summary

• Describe Exadata Storage Server metrics, alerts and

active requests
• Identify the recommended focus areas for Exadata
Storage Server monitoring
• Describe how to monitor the recommended Exadata
le
Storage Server focus areas
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

– Monitoring Exadata Storage Servers using Enterprise

Manager Grid Control and the System Monitoring Plug-in for
– Managing Exadata Storage Server Alerts and Checking for
Undelivered Alerts
– Exadata Storage Server Monitoring and Management using
Integrated Lights Out Manager (ILOM) r a ble
e s f
tra n
on-
a n
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

In this practice you will monitor Exadata Storage Server using
metrics, alerts and active requests.
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Tecn
ic
oT e i
( t
ฺ s
ko licen
o r
ec se to
t
up use
k o
te@ this
ei Stu
m b r )
ฺco dent
a s
ฺ Guid
a
h eฺ
no n - tr
a
ns fe rab
le
Monitoring Exadata Database Machine

Database Servers
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Objectives

monitoring recommendations for Exadata Database Machine

database servers.
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Monitoring Database Servers Overview
• Essentially, monitoring an Exadata Database Machine

database server is the same as monitoring any clustered

Oracle Database server
– Current skills and practices are readily transferrable
• The lesson covers Database Machine specific differences
and recommendations with a focus on using Grid Control
• Areas considered in this lesson include: le
fe rab
– Monitoring hardware
tra ns
– Monitoring the Operating System n -
a no
– Monitoring Oracle Grid Infrastructure
) h as ฺ
– Monitoring Oracle Database
m ฺbr Guide
– Monitoring Oracle Managementฺco Agent
k nt
o ude
t e i t
@ s S
o r te thi
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
For the o T part, monitoring Exadata Database Machine database servers is the same as
most
n i c
monitoring any other clustered Oracle Database server. So the skills and practices already
T e c
used by Oracle Database Administrators are readily transferrable into a Database Machine
environment.
There are some specific differences and recommendations associated with Exadata
Database Machine and Grid Control, and these are the focus of this lesson.

Monitoring Hardware
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Integrated
o TLights Out Manager (ILOM) monitors availability and sensor state using preset
n i c
thresholds for database server hardware components such as the system motherboard,
T e c
processors, memory, power supplies, fans, and network interface controllers. The availability
and sensor state can be monitored inside Grid Control using the ILOM plug-in. The following
recommendations apply:
• There are no Exadata-specific thresholds to set for database server hardware
monitoring. Failure conditions and threshold settings for the components monitored by
ILOM are preset in ILOM and are sufficient for the necessary level of monitoring.
• To view current sensor readings, log in to the Grid Control Console and navigate to All
Metrics from the plug-in home page. To view current component status, including those
that have a Faulted status, expand the Sensor Alert section under All Metrics and review
the metrics for each sensor. Components with a Faulted status will generate an alert.
Any active alert will be visible on the home page of the ILOM target.
• Alerts generated by ILOM in a database server may be viewed using Grid Control in the
same fashion as an alert generated by any other Grid Control target.
• To view the history of alerts generated by ILOM navigate to the target home page and
expand the Sensor Alert section under All Metrics and review the observations for each
sensor. A history of each sensor state is available for up to 31 days.

Monitoring the Operating System
• Operating system monitoring is built into Grid Control using

the Host target type:

– There are no Exadata-specific configuration requirements for
general monitoring
– Thresholds can be set or changed to accommodate site-
specific requirements
– Remember to monitor disk I/O in the storage servers and not
a b le
on the database servers fe r
n s
tra
– Essential operating system alerts are generated by -Grid
n
Control based on default metric thresholds no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T server operating system is viewed in Grid Control as a Host target. The
The database
o
n i c
following recommendations apply:
T e c
• There are no Exadata-specific configuration requirements. The metrics and default
thresholds provided by Grid Control are sufficient for general monitoring.
• Thresholds may be set or changed in the Metric and Policy Settings page in Grid
Control to handle site-specific requirements.
• Remember that I/O against Oracle Database data files is directed to Exadata Storage
Servers using the InfiniBand network. Therefore, metrics on database servers relating to
disk I/O cannot be used to monitor Oracle data files. I/O relating to Oracle data files
should be monitored on Exadata Storage Servers.
• Database server operating system alerts are generated automatically by Grid Control
based on default metric thresholds for the Host target.

Monitoring Oracle Grid Infrastructure
• Grid Infrastructure monitoring is built into Grid Control

using the ASM, Listener and Cluster target types:

general monitoring
– Essential alerts are generated by Grid Control based on
a b le
default metric thresholds fe r
• tra is
Monitoring of the ASM log file for cell connectivity issues
ns
o n -
not provided by Grid Control an s
ha suchฺ as:
– Check the ASM alert log file for messages
)
m ฺ br uide
connect: ossnet: connection failed
ฺ c o tontserver G <ipaddr>,
o e
result=5 (login: sosstcpreadtry
t eik Stud failed)
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T
OracleoAutomatic Storage Management (ASM), listener, and clusterware are viewed in Grid
i c
n as ASM, Listener, and Cluster targets, respectively. The following recommendations
Control
T e c
apply:
Control to handle site-specific requirements. For example, thresholds are typically
configured to provide a warning when the amount of free space in a disk group
(USABLE_FREE_MB) falls below the required amount of free space to maintain
redundancy if a failure group is lost (REQUIRED_MIRROR_FREE_MB).
• Alerts are generated automatically by Grid Control based on default metric thresholds
for the ASM, Listener, and Cluster targets.
• Monitoring of the ASM alert log file for cell connectivity issues is currently not integrated
into Grid Control. Check the ASM alert log file for messages such as the example shown
in the slide. This check should be performed every 5 minutes.

Monitoring Oracle Database
• Monitoring of Oracle databases is built into Grid Control

using Cluster Database and Database Instance targets:

general monitoring
a b le
• Monitoring of the database instance log file for cell-tra
ns
n on
connectivity issues is not provided by Grid Control
a s
ha as:
– Check the alert log file for messages such
r ) e ฺ
b
ฺto server i d
u <ipaddr>,
connect: ossnet: connection failed o m t G
result=5 (login: sosstcpreadtry i k oฺc udefailed) n
@ te St
o r te this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T
Oracleodatabases are viewed in Grid Control as Cluster Database and Database Instance
i c
n The following recommendations apply:
targets.
T e c
• Alerts are generated automatically by Grid Control based on default metric thresholds
for the Cluster Database and Database Instance targets.
• Monitoring of the database instance alert log file for cell connectivity issues is currently
not integrated into Grid Control. Check the alert log file for messages such as the
example shown in the slide. This check should be performed every 5 minutes.

Monitoring Oracle Management Agent
• Oracle Management Agent monitoring is built into Grid

Control using the Agent target type

general monitoring
a b le
• tra for
Remember that agent availability is especially important
ns
n -
the agent hosting the monitoring plug-ins a no
) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T
OracleoManagement Agents are viewed in Grid Control as agent targets. The following
n i c
recommendations apply:
T e c
• Alerts are generated automatically by Grid Control based on default metric thresholds.
By default, Grid Control generates an alert if an agent is down or loses contact with the
management server.
• Always remember which agent is hosting the monitoring plug-ins because the ability to
monitor many Database Machine components relies on the continued availability of the
agent.

Quiz
Extensive retraining is required for existing Oracle Database

Administrators to monitor and maintain Oracle databases on

Exadata Database Machine:
a. TRUE
b. FALSE
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o bT
e c n
T

Summary
In this lesson, you should have learned how to describe the

monitoring recommendations for Exadata Database Machine

database servers.
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Monitoring the InfiniBand Network
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Objectives
After completing this lesson, you should be able to describe

and execute the commands to monitor the Exadata Database

Machine InfiniBand network.
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

InfiniBand Network Monitoring Overview
DB DB DB DB
Server 1 Server 2 Server 3 Server n
InfiniBand InfiniBand InfiniBand ... InfiniBand

Port 1 Port 2 Port 1 Port 2 Port 1 Port 2 Port 1 Port 2
le
fe rab
tr ans
n -
no a
a
h eฺ s
Port 1 Port 2
InfiniBand
Port 1 Port 2
InfiniBand
Port 1 Port 2
InfiniBand ฺ Guid
b r )
Port 1 Port 2
InfiniBand
o ntm
cStorage
Storage Storage
k o ฺServer d3 e
Storage
Server 1 Server 2 i
te Stu ... Server n
r @
te this
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
T in the slide illustrates the InfiniBand communications network contained inside
The diagram
o
i c
n Database Machine.
Exadata
T e c
Each Database Machine contains at least two InfiniBand Switches which connect the
database servers and storage servers. These are called leaf switches. A third switch, called
the spine switch, connects to both leaf switches in Half Rack and Full Rack Database
Machines. The spine switch facilitates connection of multiple racks to form a single larger
Database Machine environment.
Each server contains at least one pair of InfiniBand ports which are bonded together using
active/passive bonding. The active and passive connections are spread across both leaf
switches for load balancing purposes. In addition, the leaf switches within a rack are
connected to each other. The result is a Fat-Tree switched fabric network topology.
Monitoring of the InfiniBand network is divided into three main areas:
• InfiniBand switch monitoring
• InfiniBand port monitoring
• Monitoring of the InfiniBand fabric
The remainder of this lesson covers these three areas.

Manually Monitoring the InfiniBand Switches
• InfiniBand Switch monitoring checks for failed switch

hardware and sensors that exceed preset thresholds

• Manual monitoring is required for switches running a
switch software version prior to 1.1.3
– Example of expected output:
# ssh root@dm01sw-ib2
le
rab
root@dm01sw-ib2’s password:
fe
ns
[root@dm01sw-ib2 ~]# showunhealthy
OK - No unhealthy sensors
- tra
[root@dm01sw-ib2 ~]# checkpower
no n
PSU 0 present status: OK
s a
PSU 1 present status: OK
) a
h eฺ
ฺ Guid
b r
m
– Checks should be performed
k o ฺcoevery d e 60t to 120 seconds
n
– Use env_test to gather t ei more S t uinformation if required
r @
te thi s
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
T of the InfiniBand switches provided with Oracle Exadata Database Machine
Monitoring
o
i c
n checking for failed hardware components and sensors that exceed preset thresholds
involves
e c
T the switch.
on
InfiniBand switches running switch software versions prior to version 1.1.3 do not have SNMP
support. On these switches, monitoring can be performed by logging into the switch as the
root user and running the showunhealthy and checkpower commands. An example is
shown in the slide. It is recommended that the checks should be run every 60 to 120 seconds.
In practical terms this means creating a script to perform the check and using an automated
capability (such as the User Defined Metrics capability inside Grid Control) to run it
repeatedly.
Note that showunhealthy will indicate no unhealthy sensors even if a power supply is
offline, so they must be checked separately.
If showunhealthy or checkpower indicate any issues, run the env_test command to get
more detailed status information for all switch sensors.

Monitoring the InfiniBand Switches
with Grid Control
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Commencing
o T with version 1.1.3, the InfiniBand switch software supports SNMP. In
n i c
conjunction with this, the InfiniBand Switch plug-in can be used to monitor switch status from
T e c
Grid Control. A screen shot of the InfiniBand switch home page is shown in the slide.
Monitoring the InfiniBand switches using Grid Control is simply a matter of checking that the
switches are up and dealing with any alerts that are generated. When alerts are generated,
you can use the manual commands shown on the previous page to gather additional
information.
Note that switches with switch software versions prior to version 1.1.3 can be upgraded to
current software versions however such an upgrade may also require associated upgrades to
other Database Machine components. Refer to My Oracle Support bulletin 888828.1 for
detailed version compatibility information relating to Exadata Database Machine.

Monitoring the InfiniBand Switch Ports
• Check for switch port errors with the following command:

# ibqueryerrors.pl –s RcvSwRelayErrors,RcvRemotePhysErrors,XmtDiscards,
XmtConstraintErrors,RcvConstraintErrors,ExcBufOverrunErrors,VL15Dropped
– Compare output with previous results

— SymbolErrors or RcvErrors or LinkIntegrityErrors
should not increase without LinkDowned increasing
– Check can be executed from any database server or
r a ble
InfiniBand switch
n s fe
a
n-tr
– Check should be performed every 60 to 120 seconds
no
a
) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
InfiniBand
o Tport monitoring is required to check the health of InfiniBand network ports and
n i c
interfaces. The recommended procedure is to manually execute the ibqueryerrors.pl
T e c
command as shown in the slide.
A single invocation of the ibqueryerrors.pl command shown in the slide can be used to
report switch port errors across all the InfiniBand switches. The output of the command should
be compared with previously gathered output. When all is functioning correctly, the counts for
SymbolErrors or RcvErrors or LinkIntegrityErrors should not increase without
LinkDowned increasing. The check can be executed from any database server or InfiniBand
switch. It is recommended that the check should be performed every 60 to 120 seconds.

Monitoring the InfiniBand Ports
on Database Machine Servers
• InfiniBand port monitoring is automatically performed on
Exadata Storage Servers

• Manually monitor database server ports with the following
commands:
# ibstatus
# perfquery
# ifconfig
le
# ping <Remote InfiniBand Hostname>
fe rab
# rds-ping <Remote InfiniBand Hostname>
ns tra
o n -
– Automatic monitoring with Grid Control can be n performed
using User Defined Metric (UDM) scriptsawhich
a
s are available
r ) h e ฺ
from My Oracle Support note 1110675.1 b
ฺ Gu i d
o m
k o ฺc dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
InfiniBand
o Tport monitoring on storage servers is performed by MS. No additional InfiniBand
n i c
monitoring is required on storage servers. If an InfiniBand port is not functioning correctly, MS
T e c
creates an alert, and delivers the alert Grid Control. Alert messages typically contain the
corrective action to perform to resolve the alert.
Database servers have no built-in InfiniBand monitoring. The following list describes the
recommended checks for each database server. Run the checks every 60 to 120 seconds.
• ibstatus: Check that every port shows the following output:
state: 4: ACTIVE
phys state 5: LinkUp
rate: 40 Gb/sec (4X QDR)
• perfquery: Compare output with previously gathered results. SymbolErrors or
RcvErrors or LinkIntegrityErrors should not increase without LinkDowned
increasing.
• ifconfig: Check that the InfiniBand interfaces (bondib0, ib0 and ib1) are UP.
• ping and rds-ping: Check for connectivity to all database servers and storage
servers over the InfiniBand network.
The checks in the above list can be automated in Grid Control using User Defined Metric
(UDM) scripts which are available from My Oracle Support note 1110675.1.

Monitoring the InfiniBand Fabric:
Subnet Manager Master Location
• The following checks are recommended once per day:
– Check that the InfiniBand Subnet Manager (SM) master is

located on one of the InfiniBand switches
— Example:
# sminfo
sminfo: sm lid 1 sm guid 0x21283a8516a0a0, activity count 933330
priority 5 state 3 SMINFO_MASTER
le
rab
# ibswitches
Switch : 0x0021283a8516a0a0 ports 36 "Sun DCS 36 QDR switch dm01sw-ib1"
fe
enhanced port 0 lid 1 lmc 0
tra ns
n
Switch : 0x0021283a8983a0a0 ports 36 "Sun DCS 36 QDR switch dm01sw-ib2" -
enhanced port 0 lid 4 lmc 0 no a
Switch : 0x0021283a89bda0a0 ports 36 "Sun DCS 36 QDR switch dm01sw-ib3"
a
h eฺ s
enhanced port 0 lid 3 lmc 0
b r
ฺ Gcheck) id
– For networks containing spine m
switches,
u that SM
o ฺ co ent
master is running on aespine
t ik switch t ud
S
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Daily, oit isTrecommended to check that the Subnet Manager (SM) master is located on one of
i c
n InfiniBand switches. From one of the database servers or InfiniBand switches, execute the
the
T e c
sminfo and ibswitches commands as shown in the slide. Determine the hostname for the
SM master location by matching the guid reported by sminfo with the output from
ibswitches. Alternatively, the getmaster command can be used, however note that
getmaster is only available on the InfiniBand switches, not the database servers.
In a Half Rack, Full Rack or multi-rack Database Machine environment, SM master should run
on a spine switch. A spine switch is a switch that only has other switches connected to it. The
following command identifies spine switches.
ibnetdiscover -p | awk ‘
/^SW +[0-9]+ +[0-9]+ +0x[0-9\
a-e]+ +[0-9]+x .DR - [SW|CA].*/ {
if (spine[$4]==””) spine[$4]=”yes”
if ($8 == “CA”) spine[$4]=”no”
}
END {
for (val in spine)
if (spine[val]==”yes”)
print val
}
‘

Monitoring the InfiniBand Fabric:
Network Topology and Link Status
• Use verify-topology to check the InfiniBand fabric
– Example of expected output:

# /opt/oracle.SupportTools/ibdiagtools/verify-topology
[ DB Machine Infiniband Cabling Topology Verification Tool ]

[Version 11.2.1.3.b]
Looking at 1 rack(s).....
le
rab
Spine switch check: Are any Exadata nodes connected ..............[SUCCESS]
Spine switch check: Any inter spine switch connections............[SUCCESS]
fe
Spine switch check: Correct number of spine-leaf links............[SUCCESS]
tr ans
n -
Leaf switch check: Inter-leaf link check..........................[SUCCESS]
no
Leaf switch check: Correct number of leaf-spine connections.......[SUCCESS]
a
a
h eฺ s
Check if all hosts have 2 CAs to different switches...............[SUCCESS]
Leaf switch check: cardinality and even distribution..............[SUCCESS]
b
ฺ G r ) id
• Use iblinkinfo.pl –Rl to m
monitor u
link status
o ฺ co ent
eik Stugathered
– Compare output with tpreviously d output
r @
te thi s
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
Tto checking the SM master, it is recommended to monitor the InfiniBand network
In addition
o
i c
ecn
topology and the state of all the InfiniBand network links.
T Once per day, it is recommended to execute the verify-topology command from one of
the Database Machine database servers. verify-topology conducts a series of topology-
level tests to verify that the expected storage network links are cabled correctly. An example
of the expected output for verify-topology is shown in the slide.
To check the state of all the InfiniBand network links, run the following command as root on
one of the Database Machine database servers:
# iblinkinfo.pl –Rl
Compare the resulting output with previously gathered output from a time when the InfiniBand
network was in a known good state. It is recommended to run this check frequently (every 1 to
5 minutes).

Quiz
Which of the following is true?

a. By default, InfiniBand port monitoring is automatically

performed by MS on Exadata Storage Servers
b. By default, InfiniBand port monitoring on database servers
is performed by Grid Control using User Defined Metrics
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o aT
e c n Defined Metrics can be used to monitor the InfiniBand ports on database servers,
User
T however they are not provided inside Grid Control by default. The User Defined Metrics must
be defined for each database server.

Summary
In this lesson, you should have learned how to describe and

execute the commands to monitor the Exadata Database

Machine InfiniBand network.
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

• Lesson Demonstration
– Monitoring the Database Machine InfiniBand network

– Oracle Database Machine Monitoring Best Practices
– Database Machine and Exadata Storage Server 11g Release 2
(11.2) Supported Versions
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Monitoring Other Exadata Database Machine

Components
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Objectives
After completing this lesson, you should be able to monitor the

following additional Exadata Database Machine components:

• Cisco Catalyst Ethernet Switch
• Sun Power Distribution Units
• Avocent MergePoint Unity KVM Switch
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Monitoring the Cisco Catalyst Ethernet Switch
le
fe rab
tr a ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The Cisco
o TCatalyst Ethernet switch is not in the critical data path. It does not participate in
n i c
connecting application clients to the database or in connecting database servers to the
T e c
storage servers. However, monitoring and administrative traffic does depend on the
availability of the Cisco switch.
The primary goal of monitoring the Cisco switch is to identify hardware component failure and
environmental conditions that can lead to switch malfunction. The switch monitors availability
and sensor thresholds for its hardware components.
A Grid Control plug-in facilitates easy monitoring of the Cisco switch. An example of the Cisco
switch home page is shown in the slide. The metrics and default thresholds provided by the
plug-in are sufficient for the level of monitoring necessary to ensure switch availability.
Thresholds may be changed or set in the Metric and Policy Settings page to handle site-
specific requirements. Alerts within the switch are reported using an SNMP trap which can be
received and displayed by Grid Control.
Manual monitoring of the Cisco switch can be achieved using a command-line interface that is
built into the Cisco switch. For further information, consult the documentation for Cisco
Catalyst 4500 Series switches available at www.cisco.com.

Monitoring the Sun Power Distribution Units
• Each rack contains two PDUs

• Monitor PDUs to:

– Ensure continuous power supply
– Measure power consumption
• Monitoring options:
– Physical inspection of the PDUs
le
– Remote monitoring using the
fe rab
Grid Control plug-in or other
tra ns
SNMP manager n -
Appropriate threshold settings inside
no a
—
each PDU are required to facilitate ) a

h eฺ s
remote monitoring ฺbr m G u id
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Every o T Database Machine rack contains two Power Distribution Units (PDUs). The
Exadata
i c
n provide a redundant power supply capability which helps to maintain high availability of
PDUs
T e c
the system. The primary goal of monitoring the PDUs is to ensure that a continual power
supply is maintained. PDU monitoring can also measure power usage for data center
management purposes.
On each PDU, the PDU metering unit enables you to monitor the current being used by
equipment connected in the Oracle Exadata Database Machine rack. You can monitor the
power usage in person by viewing the LCD screen on the PDU or remotely from a system on
the network. A picture of the PDU power consumption monitor is shown in the slide.
To enable remote monitoring, a Grid Control plug-in is supplied. The plug-in shows the status
of the PDU along with any alerts generated by the PDU.
Note that appropriate thresholds must be configured inside each PDU to automatically
generate alerts indicating abnormal power consumption. The specific threshold values
depend on several factors including the model of the Database Machine, the size of the
Database Machine and the power supply voltage and type. The Oracle Exadata Database
Machine Owner's Guide contains the recommended threshold values for each situation.

Monitoring the Avocent MergePoint Unity
KVM Switch
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The KVM o Tswitch provides remote console access to the Exadata Storage Servers and
n i c
database servers in Database Machine. Its primary purpose is to provide keyboard, video,
T e c
and mouse control to facilitate various administration tasks. The KVM switch is monitored
mainly to ensure it is available.
The KVM switch can be monitored using the Exadata Avocent MergePoint Unity Switch plug-
in. An example of a KVM switch home page is shown in the slide. The metrics and default
thresholds provided by the plug-in are sufficient for the level of monitoring necessary to
ensure switch availability. Alerts within the switch are reported using an SNMP trap which can
be received and displayed by Grid Control.
Note that no KVM switch exists on Exadata Database Machine X2-8. On this model, server
console functionality can be accessed using ILOM.

Quiz
It is most important to monitor which of the following

components in order to maintain the overall availability of

Database Machine?
a. Cisco Ethernet Switch
b. Sun Power Distribution Units
c. Avocent MergePoint Unity KVM Switch
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o bT
e c n
T

Summary
In this lesson, you should have learned how to monitor the

following additional Exadata Database Machine components:

• Cisco Catalyst Ethernet Switch
• Sun Power Distribution Units
• Avocent MergePoint Unity KVM Switch
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

• Lesson Demonstration
– Monitoring the Cisco Catalyst Ethernet switch and the

Avocent MergePoint Unity KVM using Grid Control
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Other Useful Monitoring Tools
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Objectives

following useful tools:

• Exachk
• DiagTools
• ADRCI
• Imageinfo and Imagehistory
a b le
• OSWatcher fe r
a ns
n- t r
no
a
) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Exachk Overview
• Exachk is a utility that:

– Collects data regarding Database Machine component versions and

best practices
– Allows administrators to check their environment against supported
version levels and best practices
– Is pre-installed on new Exadata Database Machines
– Is available from My Oracle Support note 1070954.1
– Should be executed periodically as a regular part of Database
r a ble
Machine monitoring and maintenace
n s fe
– Does not alter any Database Machine configuration settings
n - tra
o
an
– Is lightweight and has minimal impact on the system
• Exachk is NOT: ) h as ฺ
– A continuous monitoring tool m ฺbr Guide
– A replacement for Enterprise k o ฺco dent
Manager
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Exachk o T data regarding key software, hardware, and firmware versions and
collects
n i c
configuration best practices specific to Oracle Database Machine.
T e c
The output assists Database Machine administrators to periodically review and cross
reference the current data for key Database Machine components against supported version
levels and recommended best practices.
Exachk is pre-installed on new Exadata Database Machines. The latest updated version of
exachk is available for download from My Oracle Support note 1070954.1. Exachk can be
executed as desired and should be executed regularly as part of the maintenance program for
an Oracle Database Machine.
Exachk is not a continuous monitoring utility and should not be considered as an replacement
for other monitoring or alerting tools such as Enterprise Manager or ILOM.
Exachk is non-intrusive and does not change anything on the Database Machine apart from
writing output files and some small temporary files which are deleted after they are used.
Exachk will also offer to configure SSH user equivalence if it is not configured.
Exachk is lightweight so the impact to the target machine is minimal. On a Full Rack
Database Machine X2-2 system exachk normally takes a little more than one hour to run.

Running Exachk
As the Oracle Database software owner (oracle OS user):

1. Download exachk.zip onto a database server

2. Unzip exachk.zip
$ unzip exachk.zip
– Leave the exachk script and driver files (collections.dat and
rules.dat) together in the same directory
3. Ensure the exachk script is executable
le
$ chmod +x exachk
fe rab
4. Run exachk ns tra
n -
– Follow the prompts, read and understand all the messages
a no
– Supply the requested passwords, otherwise some
h a s checks are skipped
5. Review the reports r ) d e ฺ
m b
ฺ areasGfor i
u further investigation
– Review the summary report to identify o
ฺc den t
– Review the detailed reportifor k orecommendations
u and further
t e t
information
r t e @ his S
u p o se t
c ฺs © t2012,
Copyright
t e
i k o ( ense
The slide
e thelicrecommended process for installing and running exachk.
Toutlines
o
e c nicexachk executable does not require any mandatory command line options because the
The
T tool automatically scans the Database Machine environment to gather most of the required
information. Additional information is gathered using prompts at the beginning of the exachk
session. After the administrator input is gathered, exachk can be left to run unattended.
The following usage considerations should also be noted:
• It is recommended that exachk should be staged and operated from a local filesystem
on a single database server in order to deliver the best performance.
• To maximize the number of checks that are performed, execute exachk when the Grid
Infrastructure services and at least one database are up and running.
• While exachk is a minimal impact tool, it is a best practice to execute it during times of
least load on the system.
• To avoid possible problems associated with running the tool from terminal sessions on a
network attached computer, consider running the tool using VNC so that if there is a
network interruption the tool will continue to execute to completion.
• If the execution of the tool fails for some reason, then it can be re-run from the
beginning; exachk does not resume from the point of failure.

Exachk Output
[oracle@exa1db01 exachk]$ ./exachk

<Output Truncated>
Summary report - /u01/pochome/exachk/exachk_exa1db01_052411_170143/exachk_summary.rep

Detailed report - /u01/pochome/exachk/exachk_exa1db01_052411_170143/exachk.rep
UPLOAD - /u01/pochome/exachk/exachk_exa1db01_052411_170143.zip
[oracle@exa1db01 exachk]$
Check:- WARNING => ASM variable sized extents are enabled [DBMV2]
Check:- WARNING => NIC bonding is NOT configured for public network (VIP)
Check:- WARNING => kernel parameter rp_filter is set to 1.
Check:- WARNING => Free space in root(/) filesystem is less than recommended. [DBMV2]
le
rab
Check:- WARNING => One or more network cables are not connected.
fe
Check:- WARNING => kernel parameter rp_filter is set to 1.
tr a ns
Additional information to resolve above problem
n -
-----------------------------------------------
no
a
As a consequence of having rp_filter set to 1, Interconnect packets may potentially be
blocked/discarded.
a
h eฺ s
To fix this problem, use following MOS note.
ฺ Guid
b r )
TO REVIEW COLLECTED DATA :-
m
k o ฺco dent
/u01/pochome/exachk/exachk_exa1db01_052411_170143/9AA08EB2573A36C6E040E50A1EC02BD9_exa1db01
ei Stu
_report.out
t
1 Note: 1286796.1 - rp_filter for bonded private interconnects and Linux Kernel 2.6.31+ -
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Exachk o T files are written to a time stamped directory under the location where exachk is
output
n i c
installed. The main output files are:
T e c
• exachk_summary.rep which contains a summary of information, warnings and failure
notices for the exachk checks.
• exachk.rep which contains more detailed information for all of the entries in the
summary report. The entries in this report typically describe the issues in detail and
provide recommendations for remediation. A link to a note in My Oracle Support may
also be included to provide further information, especially if the recommendations are
frequently changing or they are not straight-forward.
Example output from both of the main exachk reports is shown in the slide.
Numerous additional log files are written which contain raw output from the exachk checks.
These files may be referenced in the exachk.rep report as shown in the example output on
the slide.
A zip file containing all the exachk output files is also written to the directory containing the
exachk executable.

DiagTools Overview
• Diagtools.zip contains two scripts which collect trace,

log and configuration information:

– DbmCheck.sh collects general configuration information
about all of the Exadata Storage Servers
– diagget.sh collects Oracle software log and trace files
plus OS information from all of the servers
• The scripts are designed to provide a consolidated
r a ble
package of information for use by Oracle Support
n s fe
• Diagtools.zip is available from My Oracle Supportn - tra note
735323.1 a no
•
as
See demonstration Using DiagToolsr) h deฺ
m ฺb Gui
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
Diagtools.zip is an archive that contains two scripts which collect trace files, log files and
i c
n configuration information from Exadata Database Machine:
other
T e c
• DbmCheck.sh connects to all the storage cells and collects general configuration
information such as information about cell disks, grid disks and flash cache.
• diagget.sh connects to all the Database Machine servers and collects log and trace
files related to Oracle clusterware, ASM and database. It also collects various operating
system details including information generated by OSWatcher.
The scripts are primarily designed to assist administrators to provide a consolidated package
of information about a Database Machine environment which can be used by Oracle Support
technicians to assist in resolving support requests.
DiagTools.zip is available from My Oracle Support note 735323.1. The scripts rely on the
group files (files listing the hostnames of all the Database Machine servers) generated during
initial deployment of Database Machine. The scripts should be installed at
/opt/oracle.SupportTools/onecommand on the first database server. The diagnostic
information collected by the scripts is written to date stamped directories under
/opt/oracle.SupportTools/onecommand/diagfiles.
An example of how to use the DiagTools scripts is provided in the demonstration entitled
Using DiagTools.

Using ADRCI on Exadata Storage Servers
• Exadata Storage Server software uses the Automatic

Diagnostic Repository (ADR) structure to manage

diagnostic data
• Using ADRCI on an Exadata Storage Server is essentially
the same as using it on a database server.
• The default ADR home location is:
le
/opt/oracle/cell<cell_version>/log/diag/asm/cell/<hostname>
fe rab
tr a ns
• See demonstration Using ADRCI on an Exadata -
Storage
n
Cell a no
s ha eฺ
r
ฺb Guid )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T Diagnostic Repository (ADR) is a file-based repository for diagnostic data such
The Automatic
o
i c
asntraces, dumps, alert logs, health monitor reports, and more. Exadata Storage Server
e c
T software uses an ADR structure that is essentially the same as the ADR used by Oracle
Database software. Like the Oracle Database ADR, the cell ADR can be managed using
ADRCI; the ADR command interpreter utility.
Using ADRCI on an Exadata Storage Server is essentially the same as using it on a database
server. The main difference is the location of the ADR. On Exadata Storage Servers,
/opt/oracle/cell<cell_version>/log/diag/asm/cell/<hostname> is the default
ADR home location, where <cell_version> is a string denoting the cell software version
and <hostname> is the name of the cell host. Note that each Exadata cell has a symbolic
link at /opt/oracle/cell that points to the current software version directory
(/opt/oracle/cell<cell_version>).
An example of how to use ADRCI to collect trace files for a particular incident on an Exadata
Storage Server is provided in the demonstration entitled Using ADRCI on an Exadata Storage
Cell.

Imageinfo Overview
[root@exa1cel01 ~]# imageinfo

Kernel version: 2.6.18-194.3.1.0.3.el5 #1 SMP Tue Aug 31 22:41:13 EDT 2010 x86_64
Cell version: OSS_11.2.0.3.0_LINUX.X64_101206.2
Cell rpm version: cell-11.2.2.2.0_LINUX.X64_101206.2-1
Active image version: 11.2.2.2.0.101206.2

Active image activated: 2011-02-24 12:22:44 -0800
Active image status: success
Active system partition on device: /dev/md5
Active software partition on device: /dev/md7
le
rab
In partition rollback: Impossible
fe
Cell boot usb partition: /dev/sdac1
Cell boot usb version: 11.2.2.2.0.101206.2
tra ns
n -
Inactive image version: 11.2.2.2.1.110131 no a
Inactive
Inactive
image activated: 2011-02-23 18:30:44 -0800
image status: success a
h eฺ s
Inactive system partition on device: /dev/md6
ฺ Guid
b r )
Inactive software partition on device: /dev/md8
m
k o ฺcoundefined e n t
ei Stud
Boot area has rollback archive for the version:
t
Rollback to the inactive partitions: Impossible
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Softwareo Tis initially installed on Exadata Database Machine storage servers and database
i c
n using an Exadata-specific imaging process. Exadata Storage Server patch bundles
servers
T e c
are also provided as software images.
Two utilities are provided to monitor the Database Machine server images:
• imageinfo displays information relating to the images currently installed on a server.
• imagehistory displays historical information about all the images installed on the
server.
The slide shows an example of the output displayed by imageinfo. In this example, the
command shows information associated with a storage server.
There is significantly less output associated with a database server because database
servers have no concept of active and inactive images, and they do not contain a cell boot
USB device.
The behavior and output of imageinfo can be modified using a series of optional
parameters. Execute imageinfo –h to obtain a list of the available options.

Imagehistory Overview
[root@exa1cel01 ~]# imagehistory

Version : 11.2.2.2.0.101206.2
Image activation date : 2011-01-12 14:56:44 -0800
Imaging mode : fresh
Imaging status : success
Version : 11.2.2.2.1.110131
Image activation date : 2011-02-23 18:30:44 -0800
Imaging mode : out of partition upgrade
Upgrade logs : /var/log/cellos/patch/rollback_20
le
rab
1102242002_11.2.2.2.0.101206.2_11.2.2.2.1.110131.tar.gz
Imaging status : success
fe
tra ns
Version : 11.2.2.2.0.101206.2
n -
Image activation date :
no
2011-02-24 12:22:44 -0800
a
Imaging mode
Imaging status
:
:
out of partition rollback
success a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
oT
imagehistory
i c displays historical information about all the images installed on the server.
e c n slide shows an example of the output displayed by imagehistory. In this example, the
The
T command shows information associated with a storage server. From the output it can be seen
that the storage server was initially imaged with storage server software version 11.2.2.2.0.
Then it was patched to version 11.2.2.2.1. And finally the patch was rolled back leaving the
server with the initial software version.
Typically the output associated with a database server would only relate to the initially
installed image because database servers are not patched using imaging techniques.
imagehistory has only 3 optional parameters:
-all shows all the available image attributes.
-help shows basic command using instructions.
-version <version> outputs information relating to the specified version only

OSWatcher Overview
• OSWatcher collects and archives operating system and

network metrics to aid in diagnosing performance issues

• OSWatcher gathers OS data on a regular basis, invoking
utilities such as vmstat, netstat and iostat
• OSWatcher runs on all Exadata Database Machine
servers with the following default settings:
– Metric snapshots are recorded every 15 seconds
r a ble
– 168 hours (7 days) of archived metric observations are sfe
maintained t r a n
n
– Installation location: /opt/oracle.oswatcher/oswon-
a s
– Archive location:
r ) ha eฺ
mฺb
/opt/oracle.oswatcher/osw/archiveuid o nt G
o ฺ c
• The OSWatcher User Guide
t e ik is tavailableu de from My Oracle
Support note 301137.1.
r t e @ his S
u p o se t
c ฺs © t2012,
Copyright
t e
i k o ( ense
OSWatcher
o lic of UNIX shell scripts designed to collect and archive operating
Teis a collection
e c nic and network metrics to aid in diagnosing performance issues. OSWatcher operates as
system
T a set of background processes on the server and gathers OS data on a regular basis,
invoking utilities such as vmstat, netstat and iostat.
By default, OSWatcher is configured to run on all Exadata Database Machine storage and
database servers with the following settings:
• Metric snapshots are recorded every 15 seconds.
• 168 hours (7 days) of archived metric observations are maintained.
OSWatcher is installed under /opt/oracle.oswatcher/osw. Metric observations are
maintained in the archive subdirectory.
The OSWatcher User Guide is available from My Oracle Support note 301137.1.

Quiz
Exachk is an alternative to using Grid Control:

a. TRUE
b. FALSE
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o bT
e c n
T

Quiz
Which utility should you use to provide Oracle Support with a

consolidated package of information about a Database

Machine environment?
a. Exachk
b. DiagTools
c. ADRCI
d. Imageinfo r a ble
n s fe
e. Imagehistory tra
o n -
f. OSWatcher an
) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o bT
e c n
T

Summary
In this lesson, you should have learned how to describe the

following useful tools:

• Exachk
• DiagTools
• ADRCI
• Imageinfo and Imagehistory
le
• OSWatcher fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

– Using DiagTools
– Using ADRCI on an Exadata Storage Cell
– Oracle Exadata Database Machine exachk or HealthCheck
– Exadata Storage Server Diagnostic Collection Guide
le
– OS Watcher User Guide
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Backup and Recovery
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Objectives

• Describe how RMAN backups are optimized using

• Describe the recommended approaches for disk-based
and tape-based backups of databases on Database
Machine.
• Describe the recommended best practices for backup and rable
e
recovery on Database Machine. nsf tra
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Backup and Recovery Overview
• Backup and recovery of databases on Database Machine

– Use RMAN
– Typical strategies:
— Disk-based backups
— Tape-based backups
— Hybrid strategy
• Backup and recovery of Database Machine software le
– Database server software fe rab
tra ns
– Exadata Storage Server software n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T focuses on the best practice recommendations for backup and recovery in
This lesson
o
n i c
conjunction with Database Machine. The slide lists the topics that are considered throughout
T e c
the lesson.

Using RMAN with Database Machine
• Using RMAN on Database Machine is essentially the same

as using RMAN elsewhere

– Same concepts
– Same commands
• Incremental backup performance is improved
– Block filtering is offloaded to Exadata Storage Server
le
– Fewer blocks need to be processed by RMAN
fe rab
– Offload processing is automatic and transparent ns tra
• Exadata Hybrid Columnar Compression can assist n -
a no to
further improve backup performance as ฺ
) h
– Reduced data size results in smaller,
m ฺbr quicker u idebackups
o ฺ co ent G
t eik Stud
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T
Oracleorecommends the use of RMAN for database backup and recovery in conjunction with
n i c
Database Machine. In essence, using RMAN with Database Machine is the same as using
T e c
RMAN without Database Machine.
To optimize the performance of incremental backups, the database can offload block filtering
to Exadata Storage Server. This optimization is only possible when taking backups using
RMAN. The offload processing is done transparently without user intervention. During offload
processing, Exadata Storage Server filters out the blocks that are not required for the
incremental backup in progress. Therefore, only the blocks that are required for the backup
are sent to the database. This can significantly improve backup performance, especially if the
proportion of changed blocks is low.
Exadata Hybrid Columnar Compression can assist to further improve backup performance.
When a database uses Exadata Hybrid Columnar Compression, then the number of blocks
included in the backup is reduced compared to the same data in a database that uses a less
space-efficient form of compression, or no compression at all. The reduced data size
associated with Exadata Hybrid Columnar Compression results in smaller and ultimately
quicker backups.

General Recommendations for RMAN
Use RMAN to backup and recover databases on Database

Machine
• Use RMAN incremental backups and block change
tracking
• Use an external RMAN recovery catalog repository
• Set DB_RECOVERY_FILE_DEST_SIZE to bound space
le
used in the Fast Recovery Area
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T recommendations apply when using RMAN in conjunction with Database
The following
o
n i c
Machine:
T e c
• Use RMAN incremental backups and block change tracking: For fast incremental
backups, enable block change tracking. Block change tracking allows RMAN to avoid
scanning blocks that have not changed, when creating incremental backups. Also, when
performing incremental backups of databases on Database Machine, additional block
inspection is offloaded from the database servers. Block change tracking provides the
greatest benefit for databases where fewer than 20% of the blocks are changed
between incremental backups. You may still benefit by using block change tracking with
change rates greater than 20%, but testing is recommended to ensure that backup times
are reduced.
• Use an external RMAN recovery catalog repository: The RMAN recovery catalog
should be hosted on a server outside Database Machine. In practice, it is common to
have a server that hosts the RMAN catalog along with other management repositories
such as the Oracle Enterprise Manager repository and the Oracle Secure Backup
catalog.

• Set DB_RECOVERY_FILE_DEST_SIZE to bound space used in the Fast Recovery
Area: The database writes archived redo log files and any additional recovery files to
the Fast Recovery Area. These include any disk backup files such as level 0 image
copies and level 1 backup sets as well as Flashback log files (if enabled). It is important
that you set the value of this parameter to less than the total free space in the disk
group, taking into account at least one disk failure, and preferably one Exadata cell
failure. Additionally, if multiple databases are sharing the Fast Recovery Area, ensure
that the sum of the space allocated to the different databases is less than the free space
in the disk group.
le
fe rab
tra ns
n -
a no
a s
h eฺ
r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

Disk Based Backup Strategy
For disk-based database backups, Oracle recommends:

• Use a Fast Recovery Area

• Perform an initial level 0 (full) backup
• Perform periodic incremental level 1 backups
• Update your level 0 backup by applying the second to last
level 1 backup
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T database backups, Oracle recommends incremental backups to a Fast
For disk-based
o
n i c
Recovery Area (FRA). The recommended strategy is outlined in the slide. Using this
T e c
approach, you can achieve a good compromise between the time and effort required during
backup and recovery operations. The approach also efficiently manages the amount of
storage required for backups.
Remember that the initial configuration procedure for Database Machine creates a disk group
for the FRA, and that the initial size of the disk group is determined by the backup method
chosen in the configuration worksheet.

Disk Based Backup Recommendations
• Fast Recovery Area (FRA) configuration:

– Default: FRA disk group striped across all available Exadata

Storage Servers along with data disk groups
— High availability with the best throughput
– Alternative: FRA disk group and data disk groups on
separate Exadata Storage Servers
— Separation of data and backups at the cost of some throughput le
• Additional RMAN recommendations: fe rab
– Instances and channels: tra ns
n -
— Initially, run RMAN using one database instance no
a and two
RMAN channels s
ha eฺ
—
r ) id
ฺb Gif urequired
Add another instance with two channels
m
— ฺco per instance
Add an additional two channels
k o ude nt if required
– Configure an Oracle Service t e i tot use as the RMAN target
@ s S
o r te thi
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T the installation and configuration procedure for Database Machine yields a disk
By default,
o
i c
ecn
group for the Fast Recovery Area (FRA) which is spread across all of the disks in all of the
T Exadata Storage Servers. In this configuration, the FRA shares each disk with disk groups
containing database data files. By default, all the disk groups are created with at least
NORMAL ASM redundancy (2-way mirroring). The result is that Database Machine can tolerate
the loss of an entire Exadata Storage Server and still maintain full data integrity.
Where customers desire a separation between database files and the FRA, the
recommended alternative is to configure the disk group containing the FRA on a dedicated set
of Exadata Storage Servers and to configure the required data disk groups on separate
storage servers. This configuration means that the I/O for a single workload can no longer
benefit from being striped across all the storage servers, however it also provides additional
protection if multiple simultaneous failures affect either the database or FRA. Multiple
simultaneous failures to both the database and FRA can still result in data loss.
Another strategy is to purchase additional high capacity Exadata Storage Servers specifically
to store the FRA. This allows your databases to leverage the full capacity of the Database
Machine storage servers. To implement this strategy, the FRA storage servers need to be
hosted in a separate rack and they need to be connected to the Database Machine storage
network using the spare ports on the Database Machine InfiniBand switches.

In addition to the general recommendations for RMAN in conjunction with Database Machine,
Oracle recommends the following configuration for disk based backups:
• Testing shows that optimal backup rates are achieved with 2 database instances and 2
to 4 RMAN channels per instance. For the 2 database instances designated as backup
servers, less than 10% CPU and less than 40% IO bandwidth are utilized. During
backup operations, I/O intensive parallel queries should avoid these designated backup
servers. Start by running RMAN on one database instance using two RMAN channels. If
greater backup throughput is required, use a second instance with two more channels.
Finally, if required add two more channels per instance.
• Configure an Oracle Service to run against specific database instances in the cluster.
The service is used by RMAN to automatically spread the backup load evenly among
the target instances offering the service.
• For incremental backups, use the RMAN BACKUP command option FILESPERSET 1.
This backup option will allow faster single file restore operations.
le
• Set the initialization parameter _file_size_increase_increment=2143289344
fe rab
tra ns
to optimize the space used when incremental (level 1) backups are written to the FRA.
-
• For systems running Oracle Database release 11.2.0.1, set the initialization parameter
n
no
_backup_ksfq_bufsz=4194304 to optimize read performance. Do not set the
a
a s
initialization parameter _backup_ksfq_bufsz on systems running Oracle Database
h eฺ
release 11.2.0.2 or later. r )
ฺ Guid
b
m
_backup_ksfq_bufcnt as follows: k o ฺco dent
• For systems running Oracle Database release 11.2.0.1, set the initialization parameter
t ei Stu
- On a Quarter Rack configuration set _backup_ksfq_bufcnt=32.
o
- On a Half Rack, Full r te@or multiple
Rack, t h is rack configuration set
ฺ s up use
_backup_ksfq_bufcnt=64.
( t ec se to
e i ko licen
T
c nico
T e

Tape Based Backup Strategy
For tape based database backups, Oracle recommends:

• Use media management software that is integrated with

RMAN, like Oracle Secure Backup
• Perform periodic level 0 (full) database backups
• Perform more frequent cumulative level 1 backups and
also backup the Oracle Secure Backup catalog
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
For tapeo Tbased database backups, Oracle recommends the use of Oracle Secure Backup or
i c
n media management software that is integrated with RMAN. The recommended backup
other
T e c
strategy is to perform periodic (weekly) level 0 (full) database backups. In addition, more
frequent (daily) cumulative level 1 backups should be taken along with a backup of the Oracle
Secure Backup catalog if it is the media management software which is used.

Tape Based Backup Architecture
Database Machine Sun StorageTek

SL500
Oracle Secure Backup

Administration Server
(Sun Fire X4170)
Oracle Secure Backup

Media Servers
(Sun Fire X4275)
le
fe rab
InfiniBand Fiber Channel SAN
tra ns
Network n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The slide
o Tillustrates the recommended reference architecture for a tape based backup
n i c
configuration using Oracle Secure Backup. The key recommendations associated with the
T e c
configuration follow:
• Connect the media servers to Database Machine using the high-performance InfiniBand
network. The InfiniBand connection to each media server should be bonded for high
availability. Alternatively, you can use a Gigabit Ethernet network between the media
servers and Database Machine, however if you use this configuration be aware that the
network will likely be the constraining factor for backup and recovery performance.
• Consider the bandwidth capability of the InfiniBand connection from the Database
Machine to the media server, compared to the SAN links between the media server and
the tape library. Testing shows that data rates of approximately 3 GB per second can be
sustained over an InfiniBand link. Hence you need to provide the appropriate number of
SAN links and configure them to deliver the required throughput.
• Typically, backup performance is limited by tape drive throughput. Backup performance
scales when you add more tape drives and RMAN channels. Allocate a sufficient
number of tape drives so the media servers can achieve their maximum backup and
restore rates. Add tape drives until the bandwidth of the media servers is saturated.
• Start with at least two media servers. Add media servers if you have enough tape drives
to keep them busy without saturating Database Machine resources.

Tape Based Backup Recommendations
• Media server to Database Machine network configuration

recommendation:
– Use InfiniBand for the best backup rates:
— Configure bonding of the media server InfiniBand interfaces
— Update OpenFabrics Enterprise Distribution on the media server
— Configure IP over InfiniBand connected mode for best
performance
Set the message transfer unit (MTU) size to 65520 for the r a ble
—
InfiniBand interface n s fe
— Configure the media management software to useothe n -tra
InfiniBand network an
as
• r) h deฺ
Media server SAN configuration recommendation:
ฺb Gui
– Configure persistent bindings form
co ent devices
tape
o ฺ
t eik Stud
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T can be directly connected to the Database Machine InfiniBand fabric by adding
Mediaoservers
i c
annInfiniBand Quad Data Rate (QDR) host channel adapter (HCA) to the media server. Use
e c
T InfiniBand for the best backup rates, especially for large databases that require fast backup
rates and low CPU overhead.
Follow these recommendations:
• For high availability, connect the HCA to two different Database Machine InfiniBand leaf
switches to eliminate the switch as a single point of failure. This provides transparent
failover if connectivity is lost to one of the ports. Configure bonding of the media server
InfiniBand (IB) interfaces. Use Active-Passive bonding.
• You must use an OpenFabrics Enterprise Distribution (OFED) version on the media
server that is compatible with the version found in Database Machine. For details, refer
to My Oracle Support bulletin 888828.1.
• Configure the InfiniBand interface to use IP over InfiniBand connected mode for best
performance. On Linux, edit the /etc/ofed/openib.conf file so that it contains the
entry SET_IPOIB_CM=yes. Reboot the server to enable the setting. To verify the
setting, check the contents of /sys/class/net/ib0/mode and
/sys/class/net/ib1/mode. Both files should contain the entry connected.

• Set the message transfer unit (MTU) size to 65520 for the InfiniBand interface. On
Linux, edit the /etc/sysconfig/network-scripts/ifcfg-ib* and
/etc/sysconfig/network-scripts/ifcfg-bond0 files so that they contain the
entry MTU=65520. Verify the MTU setting by examining the output of the ifconfig
command for the InfiniBand interfaces.
• To direct the backup and restore traffic over the InfiniBand network, configure the media
management software to favor InfiniBand. Note that each media management software
type has its own method of enabling this configuration. For example, Oracle Secure
Backup has the concept of a preferred network interface, which can be set on the media
server for a specific list of clients. Other media management software may require this
configuration to be defined when the software is installed. See the media management
software for information about how to direct traffic over a particular network.
For SAN attached tape devices, configure persistent bindings so the device address does not
change. If the device address changes, the media servers cannot access the device unless
le
important that your environment maintains consistent device addresses. fe rab
you update the device configuration within the media server software. Therefore, it is very
tra ns
Persistent bindings are a part of the SAN infrastructure setup. Typically, persistent bindings
n -
no
are configured through the HBA or the operating system. The configuration steps will vary by
a
a s
platform and vendor. See My Oracle Support note 971386.1 for an example of creating
h eฺ
persistent bindings for device attachments. r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

Connecting the Media Server Using Ethernet
• Ethernet can be used if throughput is sufficient:

– GigE: Expect up to 120 MB/sec from each interface

– 10gigE: Expect up to 1 GB/sec from each interface
• Recommendations:
– Use a dedicated backup network:
— Configure dedicated network interfaces on each Database
Machine database server le
– Use bonded network interfaces: fe rab
tr a ns
— Configure LACP for maximum throughput: n -
— no
Availability still maintained if one link is lost a
— a s
Configuration required on media server, network switch and
h eฺ
database servers
ฺ Guid
b r )
m
— Use Active-Passive bondingcfor
k o ฺ o high e n
d server
t
availability otherwise:
— Configure database e
t i and
servers
t umedia
S
—
o r te@configuration
No specific switch
t h is required
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T you can use a Gigabit Ethernet (GigE) or 10 Gigabit Ethernet (10gigE) network
Alternatively,
o
i c
n Database Machine and the media servers if throughput is sufficient.
between
T e c
Using a dedicated backup network eliminates any impact on the client access network. Often,
a dedicated backup network is already in place. The maximum throughput with the GigE
network is approximately 120 MB/sec for each interface. For 10gigE, the maximum
throughput is approximately 1 GB/sec for each interface. Hence a Full Rack Exadata
Database Machine X2-2 can achieve throughput up to 960 MB/sec using GigE, or 8 GB/sec
using 10 GigE, with a single dedicated network interface on each database server.
Higher throughput can be achieved using Link Aggregation Control Protocol (LACP). LACP
enables a bonded network interface to use both network channels simultaneously. To use
LACP you must configure a bonded network interface on each database server and set the
BONDING_OPTS parameter to mode=4 in the associated bonding configuration file
(ifcfg-bond1 for example). You must also configure LACP on the associated network
switch and on the media server network interfaces. If LACP is not used, use Active-Passive
bonded network interfaces to provide high availability in case of network interface failure.
Be aware that the ability to use a dedicated, bonded network interface for backup and
recovery depends on the rest of the Database Machine configuration. For example, there may
not be sufficient network interfaces available if bonding is also used for the client access
network. In this case, customers must balance the competing requirements.

Tape Based Backup Recommendations
• Run RMAN across all the available database instances

– Create a Database Service that runs across the cluster

$ srvctl add service –d <dbname> -s <service name>
-r <instance1>, ... ,<instancen>
– Use the service name and SCAN address to connect RMAN
$ rman target sys/<passwd>@<scan_address>/<service_name>
catalog ...
le
• Allocate one RMAN channel per tape drive fe rab
ns
• -tra
Configure IORM and DBRM to control resourceoallocation
n
between backups and application workloadsa n
s ha eฺ
r
ฺb Guid )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Further T
orecommendations:
i c
n• For your RAC databases, configure RMAN to run across all the database instances. It is
T e c
recommended that you create a Database Service to run across the RAC cluster. Then
when running RMAN, use the service name and SCAN address in the connect string for
the RMAN target as shown in the slide.
• Allocate one RMAN channel per tape drive. A single RMAN channel in Database
Machine can stream data at a rate of approximately 750 MB/sec from Database
Machine to the media server. Typical tape drive backup rates are between 100 MB/sec
and 240 MB/sec, depending on the drive type and compression options. Note that tape
drive compression becomes less effective when backing up tables that are compressed
at the database level.
• If Database Machine resources must be prioritized between application workloads and
backups, then configure I/O Resource Manager (IORM) and Database Resource
Manager (DBRM). This is more likely to be required in cases where time-consuming
backups to large databases must run at the same time as production application
workloads.

Hybrid Backup Strategy
• A hybrid backup strategy combines the disk-based and

tape-based backup approaches:

– Level 0 (full) database backups are stored on tape
— Relatively cheap backup media that can be stored off site
– Level 1 incremental backups are stored on disk in the FRA
— Readily available with high performance access
• Follow recommendations for both approaches le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
A hybrid
o T
backup strategy combines the disk-based and tape-based backup approaches. For
i c
ecn
example, a reasonable hybrid approach would result in level 0 (full) database backups being
T stored on tape, while the level 1 incremental backups would be stored on disk in the FRA.
The hybrid approach also combines the benefits of each storage type as indicated on the
slide. If you choose to implement a hybrid backup approach you should follow the
recommendations for disk-based backups and tape-based backups.

Restore and Recovery Recommendations
• Restore into existing data files if possible

– Restore performance is better

– Restore using all database instances
• If no existing data files are present, restore using up to 2
database instances
• Recommended number of RMAN channels:
– For disk-based restoration, use 2 to 8 RMAN channels per rab
le
database instance n s fe
n
– For tape-based restoration, set the number of RMAN - tra
channels based on the total number of tape a no
drives
s ha eฺ
r
ฺb Guid )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T rates are achieved when avoiding the overhead of initial data file allocation. If
Higherorestore
i c
n files are present prior to database restoration, do not delete the data files and perform
existing
T e c
database restoration into the existing files to take advantage of this optimization.
With pre-existing files, create a restore service across all database instances. If no existing
data files are present prior to the restore operation, create a restore service with only two
database instances. Use 2 to 8 RMAN channels per database instance for disk-based
restoration. For tape-based restoration, the number of channels should be set to the number
of tape drives.

Backup and Recovery of Database Machine
Software
• Database Server software
– Perform file system level backup and recovery

— Use your chosen file system backup management software or
Oracle Secure Backup can be used
— Copies of the Oracle Cluster Registry are automatically
maintained on the Database Server file system and should be
included in Database Server file system backups
le
• Exadata Storage Server software
fe rab
– File system level backups are not recommended
tr a ns
n -
— System areas are mirrored
no a
— Use CellCLI commands to recover if one system disk fails
a s
h ifeboth
Use the Exadata Software Rescue Procedure r ) d ฺ system
—
disks fail simultaneously m b

ฺ Gu i
o
ฺca built-in n t
The rescue procedure k o
uses e USB flash drive
ei Stud
—
t
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
It is recommended that you use file system level backup and recovery techniques for the
n i c
database servers. Use your chosen file system backup management software and
T e c
infrastructure or use Oracle Secure Backup if desired. Note that a backup of the Oracle
Cluster Registry, which also contains Voting Disk information, is automatically maintained on
the file system of the first database server and should be included in your database server file
system backups. The default location for the Oracle Cluster Registry backup is
<Grid_Home>/cdata/<Cluster_Name>, where:
• <Grid_Home> is the location of the Grid Infrastructure software as specified during the
initial configuration of Database Machine. The default <Grid_Home> location is
/u01/app/11.2.0/grid.
• <Cluster_Name> is the name of your cluster. This is the same as the DB Machine
Prefix specified during the initial configuration of Database Machine.
File system level backups are not recommended for Exadata cell software. Exadata Storage
Server maintains mirrored system areas on separate physical disks. In the rare event that
both system disks fail simultaneously, you must use the rescue functionality provided on the
CELLBOOT USB flash drive that is built into every Exadata Storage Server.

Quiz
To facilitate disk-based backups you can install additional high

capacity Exadata Storage Servers into the available space in a

Half Rack Database Machine based on high performance
disks:
a. True
b. False
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o bT
e c n Exadata Storage Servers into the same rack as a Half Rack Database Machine is only
Adding
T supported using the Half Rack to Full Rack upgrade package, and the resulting Full Rack
Database Machine can only utilize one disk type; high performance or high capacity. To add
high capacity cells to a Half Rack Database Machine using high performance cells would
require the addition of an Exadata Storage Expansion Rack or to install the high capacity cells
into a separate rack and connect them to the Database Machine storage network using the
spare ports on the Database Machine InfiniBand switches.

Quiz
How many RMAN channels should you use for tape-based

backups?
a. 2 per database instance
b. 4 per database instance
c. 1 per tape drive
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o cT
e c n
T

Summary

• Describe how RMAN backups are optimized using

• Describe the recommended approaches for disk-based
and tape-based backups of databases on Database
Machine.
• Describe the recommended best practices for backup and rable
e
recovery on Database Machine. nsf tra
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

– Backup Optimization Using RMAN and Exadata Storage

Server
– Recovery Optimization Using RMAN and Exadata Storage
Server
– Database Machine and Exadata Storage Server 11g
r a ble
Release 2 (11.2) Supported Versions
n s fe
n - tra on
– OSB - Create persistent bindings for device attachments
o
OEL an
) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Using RMAN Optimizations for Database Machine
In these practices, you will examine the backup and recovery
optimizations that are enabled when Oracle Recovery Manager

(RMAN) is used in conjunction with Exadata storage.
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Tecn
ic
oT e i
( t
ฺ s
ko licen
o r
ec se to
t
up use
k o
te@ this
ei Stu
m b r )
ฺco dent
a s
ฺ Guid
a
h eฺ
no n - tr
a
ns fe rab
le
Exadata Database Machine Maintenance

Tasks
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Objectives
After completing this lesson, you should be able to perform the

following Database Machine maintenance tasks:

• Powering Database Machine on and off
• Safely shutting down a single Exadata Storage Server
• Replacing a damaged physical disk on a cell
• Replacing a damaged flash card on a cell
a b le
• Moving all disks from one cell to another fe r
n s
• Using the Exadata cell software rescue procedure -tra
n on
s a
) a
h eฺ
ฺ Guid
b r
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Database Machine Maintenance Overview
• Maintaining Database Machine is similar to maintaining

any other clustered Oracle Database environment

• Database Machine-specific tasks outlined in this lesson:
– Powering Database Machine on and off
– Safely shutting down a single Exadata Storage Server
– Replacing a damaged physical disk on a cell
– Replacing a damaged flash card on a cell r a ble
– Moving all disks from one cell to another n s fe
– Using the Exadata cell software rescue proceduren - tra
o n
a
• Additional references:
) h as ฺ
– Oracle Exadata Database Machine r uidGuide
ฺbOwner's e
m
co ent G
– My Oracle Support o ฺ
t eik Stud
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
In many o T
respects, maintaining Exadata Database Machine is similar to maintaining any other
i c
ecn
clustered Oracle Database environment. The procedures for maintaining Oracle clusterware,
T ASM and RAC are essentially the same on Database Machine as they are on other platforms,
the main difference being the references to the Exadata cell objects.
This lesson focuses on a series of Database Machine-specific maintenance tasks that
administrators are most likely to encounter. Additional less-common maintenance tasks are
also documented in the Oracle Exadata Database Machine Owner's Guide. Administrators
should also consult My Oracle Support for notes on other maintenance issues.
Note: Patching guidelines for Exadata Database Machine are considered in a separate
lesson later in the course.

Powering Database Machine Off and On
• Power Off sequence:

1. Database servers
# <Grid_Home>/bin/crsctl stop cluster
# shutdown -h -y now
— Ensure all database servers are shut down before proceeding

2. Exadata Storage Servers
# shutdown -h -y now
le
— Ensure all storage servers are shut down before proceeding
fe rab
3. Rack, including network switches ans- tr
• Power On sequence: no n
1. Rack, including network switches s a
) a
h eฺ
— Apply power for a few minutes before proceeding
ฺ Guid
b r
2. Exadata Storage Servers o ntm
k o ฺcbefore e
—
t ei Stud proceeding
Check that all cells are running
3. Database servers e@ is
o r t t h
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
i c o Tlists the recommended sequence for powering Database Machine off and on.
The slide
e c n performing either sequence it is important to ensure that each step is completely

When
T finished before moving on to the next step. Failure to execute steps in the proper order may
result in Database Machine not functioning correctly.
When powering on the Exadata Storage Servers and the database servers, power can be
applied by pressing the power button on the front of each server, or by logging into the ILOM
interface for each server and issuing the reset /SYS command.
To power the rack on or off, use the switches located on the Power Distribution Units (PDUs)
which are located at the rear of the rack.

Safely Shutting Down a Single
• Safe shutdown sequence:
1. Check to make sure that shutting down the storage server will not take any
ASM disk group offline
CellCLI > LIST GRIDDISK WHERE asmdeactivationoutcome != ’Yes’
2. Make all the grid disks inactive

CellCLI> ALTER GRIDDISK ALL INACTIVE
3. Verify that all the grid disks are inactive

le
rab
CellCLI> LIST GRIDDISK WHERE STATUS != 'inactive'
4. Shut down the storage server fe
• Startup sequence: tra ns
n -
1. Start the storage server no a
— Cell services start automatically a
h eฺ s
2. Make all the grid disks active
ฺ Guid
b r )
CellCLI> ALTER GRIDDISK ALL ACTIVEom
o ฺ c ent
3. Verify that all the grid disks are
t k
ei active t ud asmmodestatus
S
te@ this
CellCLI> LIST GRIDDISK ATTRIBUTES name,
o r
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
In some o T
maintenance scenarios a single Exadata Storage Server must be shut down in
i c
ecn
isolation. For example, a hardware component, such as a flash memory card or a disk
T controller, may be indicating an intermittent fault so the storage server must be shut down to
replace the component while the rest of the system continues to support business
transactions. In these cases the desired result is that the Database Machine environment
continues to support processing activities without a substantial impact on the system users.
To safely and gracefully shut down a single Exadata Storage Server use the commands
shown in the slide. When checking to make sure that shutting down the storage server will not
take any ASM disk group offline you should expect to see no output from the first LIST
GRIDDISK command. If any output is returned, then it is not safe to take Exadata Storage
Server offline because proper Oracle ASM disk group redundancy will not be maintained.
Taking Exadata Storage Server offline when one or more grid disks are in this state will cause
ASM to dismount the affected disk group, causing the databases to shut down abruptly. In this
case you would need to analyze the situation and bring other grid disks online in order to
proceed safely.
After the Exadata Storage Server is restarted and the grid disks are reactivated, you should
check that all the grid disks return to an asmmodestatus of either ONLINE or UNUSED.

Replacing a Damaged Physical Disk
1 Determine the damaged disk.

CellCLI> LIST ALERTHISTORY -

WHERE ALERTMESSAGE LIKE "Logical drive lost.*" DETAIL
Logical drive lost. Lun:0_5. Status: normal. Physdisk: 20:5.
Celldisk on it: CD_05_cell01. Griddisks on it: data_CD_05_cell01.
The suggested action is: Refer to section Maintaining Physical Disks in
the User Guide.
le
2 Replace the
physical disk. fe rab
tr ns
a NORMAL
ALTER PHYSICALDISK 20:5 SERVICELED ON LIST PHYSICALDISK n -
no a
3 Monitor ASM to confirm the re-addition of the disk. ) h a s
r d e ฺ
b
ฺ Gu i
SQL> SELECT NAME, STATE FROM V$ASM_DISK om
k o ฺc dent
SQL> SELECT * FROM GV$ASM_OPERATION
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
i c o Ta physical
Replacing
maintenance
disk due to problem or failure is probably the most likely hardware
e c n ASM redundancy,that
using
operation
the
an Exadata Storage Server might ever require. Assuming you are
procedure to replace a problem disk is quite simple.
T
The first step requires that you identify the problem disk. This could occur in a number of
ways:
• Hardware monitoring using ILOM may report a problem disk.
• If a disk fails, an Exadata alert is generated. The alert includes specific instructions for
replacing the disk. If you have configured the system for alert notifications, the alert will
be sent to the designated email address or SNMP target. The LIST ALERTHISTORY
command could also be used to identify a failed disk.
• The LIST PHYSICALDISK command may identify a disk reporting an abnormal status.
Even if the cell is still functioning, the problem may be a precursor to a disk failure.
• The CALIBRATE command may identify a disk delivering abnormally low throughput or
IOPS. Even if the cell is still functioning, a single bad physical disk can degrade the
performance of other good disks so you may decide to replace the identified disk. Note
that running CALIBRATE at the same time as the cell is active will impact performance.
You can use the ALTER PHYSICALDISK command to light a service LED which assists in
correctly translating the disk name to the corresponding physical disk location.

When a disk fails, the Oracle ASM disks associated with the grid disks on the physical disk
are automatically dropped with the FORCE option, and an Oracle ASM rebalance follows to
restore the data redundancy. This process is known as proactive disk quarantine.
If you wish to replace a disk that is delivering poor performance, but which has not been
isolated by proactive disk quarantine, you must manually drop the associated grid disks using
the ALTER DISKGROUP ... DROP DISK command.
After you have identified the problem disk, you can replace it. When you remove the disk, you
will get an alert. When you replace a physical disk, the disk must be acknowledged by the
RAID controller before it can be used. This does not take a long time, and you can use the
LIST PHYSICALDISK command to monitor the status until it returns to NORMAL.
The grid disks and cell disks that existed on the previous disk in the slot will be automatically
re-created on the new disk. If these grid disks were part of an Oracle ASM disk group with
NORMAL or HIGH redundancy, they will be added back to the disk group and the data will be
rebalanced based on disk group redundancy and the asm_power_limit parameter.
a b le
Re-creating the ASM disk and rebalancing the data may take some time to complete.fYou r
e can
monitor the progress of these operations within ASM. You can monitor the status a n
ofsthe disk
t r
as reported by V$ASM_DISK.STATE until it returns to NORMAL. You can also
rebalance progress using GV$ASM_OPERATION. n on- monitor the
s a
Review the following considerations when replacing a failed disk:
) a
h eฺ
• The disk could be dropped by ASM, and the rebalance b r
ฺ operation
u id may have been
m G
successfully run. Check the Oracle ASM alert
k o ฺcologs e t
tonconfirm this. After the failed disk is
replaced, a second rebalance is required.
t ei Stud
• The disk could be dropped, and
o r t e@ t h is operation is currently running. Check the
the rebalance
GV$ASM_OPERATION view
case the rebalance ฺ s
operation u se theif the
up to following
determine rebalance operation is still running. In this
disk replacement will be queued.
t e c t o
• The disk could
i k o e n se by ASM, and the rebalance operation failed. Check
( be dropped
T e lic
GV$ASM_OPERATION.ERROR
o
to determine why the rebalance operation failed. Monitor
c n icthe rebalance operation following the disk replacement to ensure it runs.

Te • Rebalance operations from multiple disk groups can be done on different Oracle ASM
instances in the same cluster if the physical disk being replaced contains grid disks from
multiple disk groups. Multiple rebalance operations cannot be run in parallel on just one
Oracle ASM instance. The operations will be queued for the instance.

Replacing a Damaged Flash Card
1 Determine the damaged flash card.

CellCLI> LIST PHYSICALDISK DETAIL

name: [9:0:2:0]
diskType: FlashDisk
...
slotNumber: "PCI Slot: 1; FDOM: 2"
status: critical
le
2
Power down
the cell.
3
Replace the
flash card.
4
Power up
the cell.fe rab
tra ns
n -
If the card contained a flash-based grid disk,
no a
5 monitor ASM to confirm the re-addition of the disk. a
h eฺ s
b r ) id
SQL> SELECT NAME, STATE FROM V$ASM_DISK mฺ u
SQL> SELECT * FROM GV$ASM_OPERATION ฺc
o nt G
k o e
t ei Stud
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T Storage Server is equipped with 4 PCI flash memory cards. Each card has 4
Each Exadata
o
i c
n modules (FDOMs) for a total of 16 flash modules on each cell.
flash
T e c
Identifying a damaged flash module is similar to identifying a damaged physical disk.
Hardware monitoring using ILOM or a drop in performance indicated by the CALIBRATE
command may indicate a problem. If a failed FDOM is detected, an alert is generated. The
alert message includes notice of any flash-based grid disks that were on the flash module.
As shown in the slide, a damaged flash module can also be reported using the LIST
PHYSICALDISK DETAIL command. The slotNumber attribute shows the PCI slot and the
FDOM number. In this example, the status attribute indicates a critical fault.
If there were no grid disks on the flash module, the flash module was probably being used for
Exadata Smart Flash Cache. In this case, the bad flash module results in a decreased
amount of flash memory on the cell. The performance of the cell is affected proportional to the
size of flash memory lost, but the database and applications are not at risk of failure.
Although technically the PCI slots in a Exadata Storage Server are hot-replaceable, it is
recommended to power down the cell while replacing a damaged flash card.
After replacing the card and powering up the cell, no additional steps are required to re-create
any flash-based grid disks, however you should monitor ASM to confirm the re-addition of any
flash-based grid disks and to verify that all rebalance operations complete successfully.

Moving All Disks from One Cell to Another
Original New Original New

1. Make the grid disks inactive:

CellCLI>
CellCLI> ALTER
ALTER GRIDDISK
GRIDDISK ALL INACTIVE
ALL INACTIVE
2. Back up the operating system configuration files that may

change when the new cell is booted.
3. Move the disks from the original cell to the new cell. le
– Ensure the system disks occupy the first two slots. fe rab
tra ns
4. Boot the new cell. n -
5. Restart Exadata cell services:
no a
a s
h eฺALL
CellCLI>
CellCLI> ALTER
ALTER CELL CELL
RESTART RESTART
ฺ Guid
SERVICESSERVICES
b r )
ALL
m
6. Import the cell disks:
k o ฺco dent
CellCLI>
CellCLI> IMPORT
IMPORT t eiALL StuALL
CELLDISK
CELLDISK
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
You mayT
i c o need
necessary when
to move all drives from one storage server to another server. This may be
there is a chassis-level component failure, or when troubleshooting a
e c n
hardware problem. To move the drives, perform the following steps:
T
1. If possible, use the ALTER GRIDDISK ALL INACTIVE command to make the grid
disks inactive.
2. If possible, back up /etc/hosts, /etc/modprobe.conf, and the files in
/etc/sysconfig/network and /etc/sysconfig/network-scripts. This is
mainly a precautionary step and is also useful in case you wish to move the disks back
to the original chassis.
3. Shut down the original server and move the disks to the new server.
Caution: Ensure the first two disks, which are the system disks, are in the same first two
slots. Failure to do so causes the Exadata Storage Server to function improperly.
4. Start the cell. The cell operating system will be automatically reconfigured to suit the
new server hardware.
5. Restart the cell services using ALTER CELL RESTART SERVICES ALL.
6. Import the cell disks using IMPORT CELLDISK ALL.
If you are using ASM redundancy and the procedure is completed before the amount of time
specified in the DISK_REPAIR_TIME ASM initialization parameter, then the ASM disks will be
automatically brought back online and updated with any changes made during the cell
outage.

Using the Exadata Cell Software Rescue
Procedure
• Every Exadata Storage Server is equipped with a CELLBOOT
USB flash drive to facilitate cell rescue

– Required if both system disks fail simultaneously
– Use with extreme caution
• To perform cell rescue:
1. Connect to the Exadata Storage Server using the console
2. Boot the cell, and as soon as you see the "Oracle Exadata" splash
screen, press any key on the keyboard r a ble
3. In the displayed list of boot options, select the last option, n s fe
CELL_USB_BOOT_CELLBOOT_usb_in_rescue_mode,nand - trapress
Enter a no
4. Select the rescue option, and proceed with h
) asrescue
the ฺ
b r
ฺ that i d e
5. At the end of the rescue process, ensure
o m G uthe cell boots from the
system disks oฺc uden t
6. Reconfigure the cell i k
te St
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Exadatao T
Storage Server maintains mirrored system areas on separate physical disks. If one
i c
n area becomes corrupt or unavailable, the mirrored copy is used to recover.
system
T e c
In the rare event that both system disks fail simultaneously, you must use the rescue
functionality provided on the CELLBOOT USB flash drive that is built into every Exadata
Storage Server.
It is important to note the following when using the rescue procedure:
• Use extreme caution when using this procedure, and pay attention to the prompts. The
rescue procedure can potentially rewrite some or all of the disks in the cell. If this
happens, then you can irrevocably lose the contents of those disks. Ideally, you should
use the rescue procedure only with assistance from Oracle Support Services.
• The rescue procedure does not destroy the contents of the data disks or the contents of
the data partitions on the system disks unless you explicitly choose to do so during the
rescue procedure.
• The rescue procedure restores the Exadata Storage Server software to the same
release. This includes any patches that existed on the cell as of the last successful boot.

• The following is not be restored using the rescue procedure:
- The crash kernel support rpms kernel-debuginfo-common, and kernel-
debuginfo. You will need to reinstall them. These cannot be restored due to
space limitations on the CELLBOOT USB flash drive.
- Some cell configuration details, such as alert configurations, SMTP information,
and administrator e-mail address. Note that the cell network configuration is
restored, along with SSH identities for the cell, and the root, celladmin and
cellmonitor users.
- ILOM configurations. Typically, ILOM configurations remain undamaged even in
case of Exadata software failures.
• The rescue procedure does not examine or reconstruct data disks or data partitions on
the system disks. If you have data corruption on the grid disks, then do not use the
rescue procedure. Instead use the database backup and recovery procedures.
The following rescue options are available for the rescue procedure: le
• Partial reconstruction recovery: During partial reconstruction recovery, the rescuefe rab
a ns
process re-creates partitions on the system disks and checks the disks for the existence
tr
n -
of a file system. If a file system is discovered, then the process attempts to boot. If the
a no
cell boots successfully, then you use the CellCLI commands, such as LIST CELL
a s
DETAIL, to verify the cell is usable. You must also recover any data disks, as
h eฺ
r )
ฺ Guid
b
appropriate. If the boot fails, then you must use the full original build recovery option.
m
k o ฺco dent
• Full original build recovery: This option rewrites the system area of the system disks to
ei Stu
restore the Exadata software. It also allows you to erase any data on the data disks, and
t
o r te@ this
data partitions on the system disks.
ฺ s up use
• Re-creation of the CELLBOOT USB flash drive: This option is used to make a copy of
the CELLBOOT USB flash drive.
( t e c e to
To perform a rescue
e i k o using
c e nsthe CELLBOOT USB flash drive:
T to Exadata
1. Connect
o li using the console.
e c n2.icBoot the cell, and as soon as you see the "Oracle Exadata" splash screen, press any
T key on the keyboard. The splash screen remains visible for only 5 seconds.
3. In the displayed list of boot options, scroll down to the last option,
CELL_USB_BOOT_CELLBOOT_usb_in_rescue_mode, and press Enter.
4. Select the rescue option, and proceed with the rescue.
5. When prompted at the end of the rescue process proceed as follows:
a. Choose to enter a shell. Do not choose the reboot option at this point.
b. Run the reboot command from the shell.
c. During reboot, but before you see the "Oracle Exadata" splash screen, press F8 to
access the boot device selection menu.
d. Select the RAID controller as the boot device.
6. After a successful rescue, you must reconfigure the cell to return it to the pre-failure
configuration, and reinstall the kernel-debuginfo and kernel-debuginfo-
common rpms to use crash kernel support. If you chose to preserve the data when
prompted by the rescue procedure, then import the cell disks. If you chose not to
preserve the data, then you should create new cell disks, and grid disks.

Quiz
When shutting down an Exadata Database Machine, the

Exadata Storage Servers must be shut down first:

a. True
b. False
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o bT
e c n
T

Quiz
An Exadata Storage Server should be shut down to replace

failed hardware components other than hard disk drives:

a. True
b. False
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o aT
e c n the flash memory cards inside Exadata Storage Server are hot-swappable, Oracle
While
T recommends that cells should be shut down to replace hardware components inside the
chassis.

Quiz
If an Exadata Storage Server disk fails, which of the following

are true?
a. The associated ASM grid disks are automatically dropped
and an ASM rebalance occurs in order to quickly restore
redundancy
b. The disk may be replaced without shutting down the
storage server
r a ble
c. The storage server must be shut down to replace the disk n s fe
n - tra
d. Multiple ASM instances can participate in the rebalance
a no
operation of a single disk group as
r ) h eฺ
m ฺ Guid
b
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o a,T b
e c n
T

Summary
In this lesson, you should have learned how to perform the

following Database Machine maintenance tasks:

• Powering Database Machine on and off
• Safely shutting down a single Exadata Storage Server
• Replacing a damaged physical disk on a cell
• Replacing a damaged flash card on a cell
a b le
• Moving all disks from one cell to another fe r
n s
• Using the Exadata cell software rescue procedure -tra
n on
s a
) a
h eฺ
ฺ Guid
b r
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Tecn
ic
oT e i
( t
ฺ s
ko licen
o r
ec se to
t
up use
k o
te@ this
ei Stu
m b r )
ฺco dent
a s
ฺ Guid
a
h eฺ
no n - tr
a
ns fe rab
le
Patching Exadata Database Machine
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Objectives

• Describe how software in maintained on different

Database Machine components
• Locate recommended patches for Database Machine
• Describe the recommended patching process for Database
Machine
• Describe the characteristics of an effective test system erab
le
n s f
- tra
n on
s a
) a
h eฺ
ฺ Guid
b r
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Patching and Updating Overview
• There are three categories of software that must be

maintained in Database Machine:

– Software and firmware on the Exadata Storage Servers
– Software and firmware on the database servers
– Software and firmware for other components, like the
InfiniBand switches for example
• Compatibility between these different pieces of software
r a ble
needs to be maintained n s fe
• Patches and updates are rolling in nature wherevern - tra
possible a no
) h as ฺ
•
ฺ Gu e
Key information is maintained in MybrOracleidSupport
m
bulletin 888828.1 ฺco nt
i k o ude
t e S t
r @
te thi s
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
Thereo areTthree broad categories of software that must be maintained in Database Machine.
i c
n is, software and firmware in the Exadata Storage Servers, database servers and other
That
T e c
components like the InfiniBand switches. Compatibility between these different pieces of
software is vital for Database Machine to function.
Wherever possible, Database Machine patches and updates are rolling in nature, meaning
that they are applied in a manner that facilitates ongoing system availability by rolling the
patch across the environment one server at a time rather affecting all the servers at once.
Important information relating to Database Machine patches and updates is maintained in My
Oracle Support bulletin 888828.1. This bulletin is constantly updated with new information.
Oracle recommends that all Database Machine customers should sign up for an automated
alert when this bulletin changes.
The remainder of this lesson outlines additional recommendations regarding Database
Machine patching and updating.

Maintaining Exadata Storage Server Software
• Exadata Storage Server patches:

– Are complete software images

– Maintain consistency across all cell components
– Are released independently from Oracle Database patches
– Have dependencies on other software and firmware versions
• Most Exadata Storage Server patches can be applied
while databases remain up and running r a ble
• Firmware levels in Exadata Storage Server are maintainedn s fe
automatically n - tra
a no
•
) h as should
No additional software, RPMs or otherwise,
ฺ
be
ฺbr uide
installed on Exadata Storage Server
ฺ c om nt G
i k o ude
t e S t
r @
te thi s
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
Patcheso T Exadata Storage Server are provided as complete software images which contain
for
i c
n to the Linux operating system, cell server software, InfiniBand software and
updates
T e c
component firmware. This ensures that all the software and firmware components on an
Exadata Storage Server remain consistent with each other.
Exadata Storage Server patches are supplied independent of Oracle Database Server
patches. However, an Exadata Storage Server patch may require a specific Oracle Database
Server patch level, or database server firmware or operating system version. Details are
provided in the patch documentation and in My Oracle Support bulletin 888828.1.
Exadata Storage Server patches can generally be applied in a rolling manner, while the
databases continue to run. A rolling patch is performed sequentially on each storage server.
The rolling patch process relies on ASM redundancy to maintain disk group availability while
each storage server is updated and rebooted.
Patches may also include instructions for parallel installation on multiple Exadata Storage
Servers. This approach requires down time but may be preferred if a scheduled maintenance
window exists for the system.

Exadata Storage Server automatically maintains firmware within the server. Firmware levels
are periodically checked while Exadata Storage Server is running and the correct firmware is
automatically applied to components when the server reboots. Disk firmware is also
automatically updated when a disk is replaced.
Oracle recommends that no additional software, whether RPMs or otherwise, should be
installed on Exadata Storage Server.
le
fe rab
tra ns
n -
a no
a s
h eฺ
r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

Maintaining Database Server Software
• You can patch and update the database server software

as you would for an Oracle Database server outside of

Database Machine
– Oracle Database patches applied by OPatch
— Double-check compatibility of patches with Database Machine
— Oracle supplies Bundle Patches for Database Machine
— Periodic bundle of database patches recommended for le
Database Machine
fe rab
– Operating system and firmware updated using regulara
tr ns
channels o n -
n
—
s a software
Must maintain consistency with InfiniBand (OFED)
hafor database
• Check Exadata Storage Server patches r ) d e ฺ
m b
ฺ Gu i
server firmware and operating o
ฺc dent updates
system
k o
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T you can patch and update the database server software using the same
In essence,
o
i c
ecn
techniques and approaches as you would for an Oracle Database server outside of Database
T Machine.
OPatch is used to apply and manage Oracle Database patches. It is recommended that you
raise a Service Request with Oracle Support to verify the compatibility of patches with
Database Machine. In addition to regular database patches and updates, Oracle supplies
periodic Bundle Patches for Database Machine. These bundle a series of database patches
that are recommended for use in conjunction with Database Machine. Oracle recommends
that you should apply Bundle Patches as part of your patching regime.
Operating system and firmware patches should be managed using standard patching
practices. My Oracle Support bulletin 888828.1 lists current constrains and requirements for
operating system and firmware patches. The main consideration is to ensure that any patches
or updates maintain consistency with the OFED software version which underpins the
InfiniBand network.
Some Exadata Storage Server patches also include database server firmware and operating
system updates. This may occur, for example, when the firmware for the InfiniBand HCA is
updated on Exadata Storage Server and the update must also be applied to the database
servers.

Assisted Patching Using OPlan
• OPlan is a utility that provides step-by-step patching

instructions specific to the target environment

– Can create instructions for Apply and Rollback
• OPlan works in conjunction with Exadata Database
Machine recommended Bundle Patches
– Starting with 11.2.0.2 Bundle Patch 2
le
• Using OPlan:
fe rab
– As the Oracle software owner:
tra ns
— Download the Bundle Patch to a local directory on-
Set the $ORACLE_HOME environment variable a n execute:
and
—
s ha eฺ
r
ฺb Guid
$ORACLE_HOME/oplan/oplan generateApplySteps <bundle patch location> )
m
— Locate the customized patch
k o ฺcoinstallation
e n t instructions at:
t ei Stud
$ORACLE_HOME/cfgtoollogs/oplan/<TimeStamp>/InstallInstructions.txt
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
OPlanois T
a utility that facilitates the patch installation process by providing step-by-step
i c
n instructions specific to your environment. OPlan works in conjunction with Database
patching
T e c
Machine recommended Bundle Patches.
Without OPlan, administrators must read the patch documentation and translate the generic
patch installation instructions into specific commands suitable for the target environment. For
example, references to the Oracle home directory or the path to the patch directory must be
translated into the appropriate directory paths. Also, administrators may be required to choose
between a number of options and evaluate the configuration of the environment to generate
the right set of commands.
OPlan assists administrators by automatically analyzing the environment and collecting the
required configuration information. Using this information, OPlan generates a set of
instructions and commands which is customized specifically for the target configuration.
OPlan can create the instruction required to apply a patch and also the instruction required to
perform a patch rollback.
The OPlan utility can be downloaded from My Oracle Support using patch number 11846294.
OPlan supports all Exadata Database Machine recommended Bundle Patches commencing
with release 11.2.0.2 Bundle Patch 2. See My Oracle Support bulletin 1306814.1 for further
information on OPlan.

Maintaining Other Software
• Other components in Database Machine that have

software or firmware:
– InfiniBand switches
– Power Distribution Units (PDUs)
– KVM switch
– Cisco Ethernet switch
• Always refer to My Oracle Support bulletin 888828.1 for rable
e
Exadata-specific requirements nsf tr a
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Other o T Machine components that have software or firmware include:
Database
i c
n• InfiniBand switches
T e c
• Power Distribution Units (PDUs)
• Keyboard, Video and Mouse (KVM) switch
• Cisco Ethernet switch
The software and firmware on all components must always be maintained in accordance with
the guidelines published in My Oracle Support bulletin 888828.1. In particular this is critical for
the InfiniBand switches. For other components the requirements may be less specific.

Recommended Patching Process
1. Review the patch documentation (README file)

– Read and understand it all before proceeding

2. Validate the patch installation on a test system
– Run exachk before and after patch application
– Automate the patch application steps where possible
– Test the failback procedure
le
3. Validate the patch functionality on a test system
fe rab
– Verify that the patch provides the desired functionalityra ns
– Evaluate system performance o n -t
n a
4. Apply the patch in production a
h eฺ s
b r )
ฺ application id
– Run exachk before and after patch m u
ฺcobefore n G
tapplying the patch
– Make sure all cells are healthy k o e
– Evaluate system performance t ei Stud
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T are the same. This adage is true whether or not you are using Database
No twoopatches
n i c
Machine. Hence you should always follow a methodical patching process. Following is an
T e c
outline of the recommended patching process for Exadata Database Machine:
1. Review the patch README file for known issues, patch installation and deinstallation
instructions, and special notes.
2. Validate the patch installation on a proper test system:
- Run exachk to verify the environment before and after the patch application.
- Where possible, automate the steps to reduce human error.
- Test the fallback procedures in case the patch must be rolled back.
3. Validate patch functionality on the test system.
- If applying the patch to address a specific issue, verify that the patch provides the
desired functionality.
- Verify that there are no performance, availability, or operational regressions. Test
using a workload that is representative of the production workload. Compare
metrics observed during the test with baseline metrics observed in production.
Real Application Testing may be used to replay a production workload on the test
system. Automatic Workload Repository (AWR) and SQL Performance Analyzer
may be used to assess performance on the test system.

4. Apply the patch on the production system.
- Run exachk to verify the environment before and after the patch application.
- Make sure that all the Exadata Storage Servers are functioning correctly. This is
especially important because an unnoticed storage server fault that doesn't seem
to impact Database Machine operations may result in diskgroups being offlined
and databases terminating abruptly when another storage server taken offline for
patching.
- Compare metrics captured before and after patch application to evaluate system
performance. SQL Performance Analyzer may be used to assess performance
improvement or regression resulting from the patch.
See the lesson entitled Other Useful Monitoring Tools and My Oracle Support note 1070954.1
for details regarding exachk.
le
fe rab
tra ns
n -
a no
a s
h eฺ
r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

Test System Recommendations
• An effective test system:

– Is an exact replica of the production Database Machine

– Is not used for any other purpose during testing
– Contains a full copy of the production data set with identical
statistics
– Is able to mimic production transaction volumes and
concurrency
r a ble
– Is able to compare workload and performance metrics onsthe
n fe
test system with metrics collected from the production
n - tra
system
a no
– Includes the operational and availabilityha s suites used in
test
r ) d e ฺ
production ฺ b u i
ฺ c om nt G
i k o ude
t e S t
r @
te thi s
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
T list describes the recommendations for an effective Database Machine test
The following
o
n i c
environment:
T e c
• A test system should be an exact replica of the primary Exadata Database Machine
including any standby systems for high availability and disaster recovery.
• When the test system is shared for other purposes, such as development, exclusive
time should be allocated to perform patch validation.
• The test system should contain a full copy of the production data set with identical
statistics.
• A workload framework, such as Real Application Testing, should be used to mimic
production transaction volumes and concurrency.
• AWR data should be collected from the production database to compare against AWR
data collected on the test system.
• Any operational and availability test suites used in production should also be ported to
the test environment.

Quiz
Which of the following statements is a broad overview of the

recommended approach for patching and updating Database

Machine?
a. All updates for Database Machine are specific to Database
Machine, and these are the only updates that should be
applied to Database Machine
b. Use Unbreakable Linux Network (ULN) to update Linux on able
database servers and Exadata servers, and apply specific s fer
t r a n
database and Exadata updates for the remainingnsoftware
n o -
c. Use Exadata-specific update bundles for the s a Exadata
servers, and use normal database updating) a
h practices
ฺ for
ฺ b r i d e
the database servers om t G u
i k oฺc uden
@ te St
o r te this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o cT
e c n
T

Summary

• Describe how software in maintained on different

Database Machine components
• Locate recommended patches for Database Machine
• Describe the recommended patching process for Database
Machine
• Describe the characteristics of an effective test system erab
le
n s f
- tra
n on
s a
) a
h eฺ
ฺ Guid
b r
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

– Exadata Storage Server Rolling Patch Application

– Exadata Storage Server Patch Rollback
– Database Machine and Exadata Storage Server 11g
Release 2 (11.2) Supported Versions
– Oracle Software Patching with OPLAN
r a ble
e
nsf
– Oracle Exadata Database Machine exachk or HealthCheck
tra
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Exadata Database Machine Automated

Support Ecosystem
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Objectives

• Describe the Auto Service Request (ASR) function and

how it relates to Exadata Database Machine
• Describe the implementation requirements for ASR
• Describe the ASR configuration process
• Describe Oracle Configuration Manager (OCM) and how it
relates to Exadata Database Machine r a ble
e s f
tra n
on-
a n
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Auto Service Request Overview
• Automatically and securely opens service requests with

Oracle for common server hardware faults

– Minimal data is collected and transmitted
– No IP addresses are communicated
– One-way (customer to Oracle) SSL encrypted
communications are used
• Enables fast and accurate resolution of hardware faults le
fe rab
– Improved availability, less downtime
ra ns
• Can be integrated with existing monitoring toolson-t
aSNMP traps to n
– ASR manager can send SR notifications via
s
a
existing monitoring tools r) h deฺ ฺb Gui
m
• Included with hardware warranty
k o ฺco dand e ntOracle Premier
Support for Systems tei tu
@ s S
o r te thi
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T Request (ASR) is a secure, scalable, customer-installable software solution
Auto Service
o
n i c
available as a feature of your Oracle or Sun hardware warranty, and Oracle Premier Support
T e c
for Systems or valid Sun support plan. The ASR software helps to resolve problems faster by
using auto-case generation for Oracle's Sun server and storage systems when specific
hardware faults occur.
With the release of the Exadata software version 11.2.1.3.1, ASR functionality is extended to
Exadata Database Machine. ASR automatically opens service requests (SRs) with Oracle
Support when specific hardware faults occur either in the Exadata Storage Servers or the
database servers. Note that ASR is currently applicable only for hardware faults detected on
the following server components: CPUs, disk controllers, disks, flash cards, flash modules,
InfiniBand cards, memory modules, system boards, power supplied, and fans.

The ASR Process
Customer Datacenter Oracle Support Services

FRU replaced by
Field Engineer
Oracle Support
Engineer
Oracle Field Field Replaceable

Engineer Unit (FRU)
dispatched by
Support Engineer
Service
Request
Fault routed to
le
rab
occurs Support
Customer SR creation fe
Engineer
email notification
to customer tr a ns
n -Oracle Case
no
Management
System
s a
) a
h eฺ
ASR
Manager b r
ฺ Guid Service Request
ASR Service
SNMP trap sent to
ASR Manager m
Fault telemetry
co to ent (SR) created
ฺ
securely
o
eik Oracle ud
transmitted
t S t
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T in the slide illustrates the ASR process. It shows how a Service Requests (SR)
The diagram
o
isn i c
automatically opened by the ASR Manager after it receives an SNMP trap that is triggered
e c
T a server hardware fault. Within Oracle Support Services the automatic service request is
by
acknowledged via email and serviced according to normal procedures.
Customers should note that there are occasions when a SR may not be automatically filed.
This can happen due to loss of connectivity to the ASR manager, for example. Oracle
recommends that customers continue to monitor their systems for faults and engage Oracle
Support if they notice a fault but do not receive notice that a service request has been
automatically filed.

ASR Requirements
• A server capable of running ASR Manager:

– An existing ASR Manager (version 2.7 or greater) can be used to

monitor Database Machine
– A Database Machine database server can be used as the ASR
Manager (not recommended)
• Connectivity between the ASR manager and the Database
Machine management network
• HTTPS connectivity (either directly or via a proxy) from the ASR rabl
e
manager to Oracle Support n s fe
• n -
My Oracle Support account with current contact information tra for
all Database Machine assets a no
• Sun Online Account (SOA) ) h as ฺ
m ฺ br uide
• Exadata software 11.2.1.3.1 or later
o ฺ co onethe
n tGdatabase servers and
Exadata Storage Servers eik ud
@ t S t
r te thi s
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
i c o Tlists the requirements for using ASR in conjunction with Database Machine.
The slide
e c n the Hardware and Network Configuration Recommendations for ASR at

See
T http://www.oracle.com/technetwork/server-storage/asr/overview/hardware-recommendations-
330108.html for more details regarding the server requirements for the ASR Manager.
When checking for the correct Exadata software on the Database Machine database servers,
check for the existence of the following file:
/opt/oracle.cellos/compmon/exadata_mon_hw_asr.pl
If the file is missing, then update the database servers using a current Bundle Patch.
By default, configuration of ASR is performed by Oracle engineers as part of the initial
installation and configuration of Exadata Database Machine. Customers can opt-out of
configuring ASR however this is not recommended by Oracle. The next section in this lesson
outlines the ASR configuration process. This process can be used by existing Exadata
customers to configure ASR after Exadata Database Machine is already installed, or it can be
used simply as a quick reference so that customers can better understand the ASR
configuration process.

Configuring ASR Manager
1. Install ASR Manager components:

– Oracle Service Tools Bundle (STB)

– Service Tags
– SAMS Package
– ASR Package
2. Register ASR Manager:
le
– As root on the ASM Manager server, run:
fe rab
# asr register
trans
—
on-
Follow the prompts and provide requested information
n
a
– To check registration status, run:
) h as ฺ
# asr show_reg_status
m ฺbr Guide
– To test the network connectivity,
k o ฺco drun: e nt
# asr test_connection t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
i c o T a new ASR Manager server consists of two main tasks:
Configuring
e c n1. Installing the ASR Manager software components

T 2. Registering the ASR Manager
The procedure for installing the ASR Manager software components varies depending on the
ASR Manager version and the platform selected to run the ASR Manager. Consult the ASR
documentation at http://www.oracle.com/technetwork/server-
storage/asr/documentation/index.html for a detailed installation guide. A quick installation
guide entitled Oracle Exadata Quick Installation Guide for ASR can also be found at the same
location.

Configuring Exadata Database Machine
for ASR
1. Configure SNMP trap destinations:
– As root on each database server, run:

# /opt/oracle.cellos/compmon/exadata_mon_hw_asr.pl -set_snmp_subscribers \
> "(type=asr,host=<ASR Manager>,port=162,community=public,)"
– As root or celladmin on each Exadata Storage Server, run:

CellCLI> alter cell snmpsubscriber = -
> ((host='<ASR Manager>', port=162, community=public, type=ASR))
2. Verify ASR SNMP subscribers: le

– As root on each database server, run: fe rab
# /opt/oracle.cellos/compmon/exadata_mon_hw_asr.pl -get_snmp_subscribers \
tr a ns
> -type asr
n -
a Server, run:
– As root or celladmin on each Exadata Storage no
) h as ฺ
ฺbr Guide
CellCLI> list cell attributes snmpsubscriber
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T Database Machine database server and Exadata Storage Server must be
Each Exadata
o
n i c
configured to deliver SNMP traps to the ASR Manager server. Use the commands shown in
T e c
the slide to configure and verify the required SNMP trap destinations.
In the commands, any reference to <ASR Manager> should be replaced with the ASR
Manager server hostname or IP address.
Remember that the SNMP subscriber configuration is required for every Database Machine
database server and Exadata Storage Server. If you prefer, you can use the dcli utility to
perform the configuration simultaneously across multiple servers. For example, the following
command could be used to perform the SNMP subscriber configuration on all the storage
servers listed in the cell_group file:
# dcli -g cell_group -l celladmin -n "cellcli -e \
> alter cell snmpsubscriber=$\(host=<ASR Manager>, \
> port=162, community=public, type=ASR$\)"

Activating ASR Assets
• As root on the ASR Manager server:

– Activate the ASR Manager host:

# asr activate_asset -i <ASR Manager IP>
– Activate each Database Machine database server ILOM and

each Exadata Storage Server ILOM:
# asr activate_asset -i <Asset ILOM IP>
– Activate each Database Machine database server and each rabl

e
e
Exadata Storage Server: nsf tra
# asr activate_exadata -i <Asset IP> \ n -
> -h <Asset hostname> -l <Asset ILOM IP> no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The slide
o Tshows the commands required to activate the ASR assets. All commands are
n i c
performed as root on the ASR Manager server. Activating the ASR Manager host is a one-
T e c
time operation, while the other commands must be executed once for each Database
Machine database server and Exadata Storage Server.
In the commands, <ASR Manager IP> should be replaced with the ASR Manager server IP
address, <Asset ILOM IP> should be replaced with the database server or storage server
ILOM interface IP address, <Asset IP> should be replaced with the database server or
storage server management network IP address, and <Asset hostname> should be
replaced with the database server or storage server hostname.

Approve and Verify ASR Assets
• Approve Database Machine assets in My Oracle Support:

1. Click the More... tab and then Settings

2. Click Pending ASR Activations in the Settings left-hand section
3. Select a Database Machine server from the list
4. Review the ASR Activation – Asset window
5. Click Approve to complete activation
6. Repeat steps 3, 4 and 5 for until all the Database Machine assets
le
are activated
fe rab
• Verify the ASR configuration:
tr a ns
– As root on each database server, run: n -
no a
> -type asr a
h eฺ s
# /opt/oracle.cellos/compmon/exadata_mon_hw_asr.pl -validate_snmp_subscriber \
– As root or celladmin on each Exadata b r )

ฺ Storage id Server, run:
m G u
k
CellCLI> alter cell validate snmp type=ASR o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
After the
o TDatabase Machine server assets are activated in the ASR Manager, the servers are
i c
n inside My Oracle Support as Pending ASR Activations. To complete the activation
listed
T e c
process the servers must be approved in My Oracle Support using the following process:
1. On the My Oracle Support home page, click the More... tab and then Settings.
2. Click Pending ASR Activations in the Settings panel on the left side of the window .
3. Select a Database Machine server from the list.
4. Review the ASR Activation – Asset window. Update information relating to your
Database Machine environment as required.
5. Click Approve to complete activation.
6. Repeat steps 3, 4 and 5 for until all the Database Machine assets are activated.
After the assets are approved, the ASR configuration is completed and the system is enabled.
To test the configuration and verify that it functions correctly, execute the commands shown
on the slide. Each time a validation command is executed, the end-to-end ASR process is
tested and a message is sent to the Sun Online Account (SOA) email account indicating that
the ASR service inside Oracle Support received the test message. No Service Request is
opened for validation requests.

Oracle Configuration Manager Overview
Oracle Configuration Manager collects configuration

information for propagation to Oracle Support

• Benefits include:
– More efficient problem diagnosis
– Easier service request reporting
• Configuration information can be uploaded to Oracle either
automatically or manually r a ble
• Configuration information remains confidential n s fe
n - tra
• no with
Oracle recommends the use of OCM in conjunction
a
Exadata Database Machine as ) h eฺ
ฺ Guid
b r
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T
OracleoConfiguration Manager (OCM) automatically collects configuration information from
i c
n environment at regular intervals. This configuration information can be uploaded to My
your
T e c
Oracle Support. This helps Oracle to maintain up-to-date information about your environment,
diagnose support issues more efficiently, and offer consistently better support outcomes.
OCM is recommended for use in conjunction with Exadata Database Machine.
Oracle recommends that OCM is set up so that the configuration information can be
automatically collected and uploaded to the Oracle Customer Configuration Repository and
My Oracle Support. Alternatively, information can be manually uploaded by an administrator.
No business or personal information is collected and uploaded, except for contact information
which is used in the event of transmission problems. All information is kept strictly
confidential.
Note that OCM is not a function of Grid Control and is separate from the Grid Control
Configuration Management Pack.

Configuring Oracle Configuration Manager
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Like ASR,
o Tconfiguration of OCM is performed by Oracle engineers as part of the initial
n i c
installation and configuration of Exadata Database Machine. Customers can opt-out of
T e c
configuring OCM however this is not recommended by Oracle.
For customers who wish to configure OCM against an existing Database Machine, the easiest
way is to use the resources under the Collector tab in My Oracle Support. The slide show the
interface presented to uses which enables them to get started with OCM.
Note that the OCM collector is installed on the Database Machine database servers. No OCM
components are installed on the Exadata Storage Servers. Therefore select Linux x86-64 or
Oracle Solaris on x86-64 (64-bit) from the Select Platform drop-down list depending on how
your Database Machine is configured.

Quiz
Oracle recommends that the Auto Service Request Manager is

configured on one of the Exadata Database Machine servers:

a. True
b. False
le
fe rab
tr a ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o bT
e c n ASR Manager can be configured on an Exadata database server, but this is not
The
T recommended.

Quiz
ASR will create an service request with Oracle Support when

which of the following occur?

a. A disk fault in a storage server
b. An ORA-600 error in a database server
c. A power supply in a server fails
d. A hardware fault in a server fan
a b le
e. All of the above fe r
a ns
n- t r
no
a
) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o a,T c, d
e c n
T

Quiz
The Grid Control Configuration Management Pack is required

to use Oracle Configuration Manager:

a. True
b. False
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o bT
e c n Configuration Manager is not a function of Grid Control and is separate from the Grid
Oracle
T Control Configuration Management Pack.

Summary

• Describe the Auto Service Request (ASR) function and

how it relates to Exadata Database Machine
• Describe the implementation requirements for ASR
• Describe the ASR configuration process
• Describe Oracle Configuration Manager (OCM) and how it
relates to Exadata Database Machine r a ble
e s f
tra n
on-
a n
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

• Documentation and recommendations

– Oracle Auto Service Request User Documentation and

Product Qualification Information
– Hardware and Network Configuration Recommendations for
ASR
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Quality of Service Management
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Lesson Objectives

• Describe the purpose of Oracle Database Quality of

Service (QoS) Management
• Describe the benefits of using Oracle Database QoS
Management
• Describe the components of Oracle Database QoS
Management r a ble
• Describe the operation of Oracle Database QoS n s fe
Management n - tra
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

QoS Management Background
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T Oracle Database releases, you could use services for workload management and
In previous
o
i c
ecn
isolation. For example, a group of servers might be dedicated to data warehouse work, while
T another is dedicated to your sales application, a third group is used for ERP processing, and a
fourth group to a custom application. Using services, the database administrator can allocate
resources to specific workloads by manually changing the number of servers on which a
database service is allowed to run. The workloads are isolated from each other, so that
demand spikes, failures and other problems in one workload do not affect the other
workloads. The problem with this type of deployment is that each workload needs to be
separately provisioned for peak demand because resources are not shared.
You could also define services which shared resources by overlapping server allocations.
However even with this capability, you had to manually mange the server allocations and
each service was mapped to a fixed group of servers.
Starting with Oracle Database 11g, you can use server pools to logically partition a cluster
and provide workload isolation. Server pools provide a more dynamic and business focused
way of allocating resources because resource allocations are not dependant on which servers
are up. Rather, the server pool allocations dynamically adjust when servers enter and leave
the cluster to best meet the priorities defined in the server pool policy definitions.

QoS Management Overview
Define and Enable Classify and Measure

QoS Management Policy Set

Maintenance Policy
Weekend Policy
Server
Evaluate and Policy-Driven
After Hours Policy Analyze and
Pools
le
rab
Business Hours
Report Recommend
Architecture Policy
fe
ns
Performance Objectives
Performance
Classes Business Rankings
- tra
Server Pool Allocations
no n
s a
) a
h eฺ
ฺ Guid
b r
m
k o ฺco dent
t ei andSControl
Implement tu
r @
te thi s
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
Many o T
companies are consolidating and standardizing their data center computer systems. In
i c
n with this, the migration of applications to the Internet has introduced the problem of
parallel
T e c
managing demand surges that cannot be fully anticipated . In this type of environment, it is
necessary to pool resources and have management tools that can detect and resolve
bottlenecks in real time. Policy-managed server pools provide a foundation for dynamic
workload management however they can only adjust resource allocations in response to
server availability changes.
QoS Management is an automated, policy-based workload management (WLM) system that
monitors and adjusts the environment to meet business level performance objectives. Based
on resource availability and workload demands, QoS Management identifies resource
bottlenecks and provides recommendations for how to relieve them. It can make
recommendations for the system administrator to move a server from one server pool to
another, or to adjust access to CPU resources using the Database Resource Manager, in
order to satisfy the current performance objectives.
Using QoS Management enables the administrator to ensure the following:
• When sufficient resources are available to meet the demand, business level
performance objectives for each workload are met, even if the workloads change.
• When sufficient resources are not available to meet all demands, QoS Management
attempts to satisfy more critical business objectives at the expense of less critical ones.

QoS Management and
• In its initial form, QoS Management is a feature of the
Oracle Database product family:

– Introduced in Oracle Database 11g release 2
– Associated with Oracle RAC software
– Released exclusively on
– Focused on environments supporting
r a ble
multiple OLTP workloads
n s fe
– Not Exadata-specific technology
QoS Management Policy
n -
Maintenance tra
Policy
Set
– The first step along the road n o Policy

Weekend
Server
a
hPolicy-Driven
as ฺ Business
After Hours Policy
Pools
towards a broader solution Hours
ฺ b
Performance
r) Architecture
Policy
u ide
m
co ent G
Classes Business Rankings
o ฺ Server Pool Allocations
t eik Stud
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Tincarnation of QoS Management is as a feature of the Oracle Database product
The initial
o
i c
n in association with Oracle Real Application Clusters (RAC) software. It was first
family
T e c
introduced in Oracle Database 11g release 2.
The initial set of features and benefits associated with QoS Management are exclusively
available to Exadata Database Machine customers and are best suited to customers using
Database Machine predominantly as a consolidation platform for OLTP applications.
QoS Management software can operate on non-Exadata environments where Oracle
Database 11g release 2 is available. Commencing with version 11.2.0.3, a limited subset of
QoS Management functionality has been released which enables non-Exadata users to
monitor performance classes, but not generate and implement changes in response to the
currently observed workload.
In its current form, QoS Management provides a powerful database-focused capability which
represents the first step along the road towards a broader workload management solution.

QoS Management Focus
Code Development
Resource Capacity Planning

Use Configuration/Deployment
Resource Quality of Service

Wait
Management le
fe rab
tra ns
n -
no a
Application a
h eฺ s
ฺ Guid
b r )
Performance m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
QoS Management monitors the performance of each work request on a target system. By
n i c
accurately measuring the two components of performance, resource use and wait,
T e c
bottlenecks can be quickly detected and resources reallocated to relieve them, thus
preserving or restoring service levels. Changing or improving the execution time generally
requires application source code changes. QoS Management therefore only observes and
manages wait times.
QoS Management bases its decisions on observations of how long work requests spend
waiting for resources. Examples of resources that work requests might wait for include
hardware resources, such as CPU cycles, disk I/O queues, and global cache blocks.
Other waits can occur within the database, such as latches, locks, pins, and so on. While
these database waits are accounted for by QoS Management, they are not broken down by
type or managed. Minimizing unmanaged waits requires changes that QoS Management
cannot perform, such as application code changes and database schema optimizations for
example. QoS Management is still beneficial in these cases because the measurement and
notification of unmanaged waits can be used as a tool to measure the effect of application
optimization activities.

QoS Management Benefits
• Determines where additional resources are needed

• Determines if additional hardware can be added to

maintain acceptable performance
• Reduces number of critical performance outages
• Reduces time to resolve performance objective violations
• Improves system stability as the workload changes
a b le
• Helps to ensure that SLAs are met fe r
n s
• Facilitates effective sharing of hardware resources-tra
n on
s a
) a
h eฺ
ฺ Guid
b r
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Some o of T
the benefits of QoS Management include:
i c
n• By categorizing and measuring database work, QoS Management can help
T e c
administrators determine where additional resources are needed.
• QoS Management is Oracle RAC-aware, and it uses this fundamental understanding to
determine if additional hardware can be added to maintain acceptable performance.
• QoS Management helps reduce the number of critcal performance outages. By
reallocating runtime resources to the busiest business-critical applications, those
applications are less likely to suffer from a performance outage.
• QoS Management reduces the time needed to resolve performance objective violations.
Rather than requiring administrators to understand and respond to changes in
performance, much of the work can be automated. Administrators are provided with a
simple interface to review and implement the recommended changes.
• Performance stresses can often lead to system instability. By moving resources to
where they are most needed, QoS Management reduces the chance that systems will
suffer from performance stress and related instability

• QoS Management allows the administrator to define performance objectives that help to
ensure Service Level Agreements (SLAs) are being met. Once the objectives are
defined, QoS Management tracks performance and recommends changes if the SLAs
are not being met.
• As resource needs change, QoS Management can reallocate hardware resources to
ensure applications make more effective used of those resources. Resources can be
removed from applications that no longer require them, and added to an application that
is suffering from performance stress.
le
fe rab
tra ns
n -
a no
a s
h eฺ
r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

QoS Management Functional Overview
QoS Management works with Oracle RAC and Oracle

Clusterware to:
• Manage database server CPU resources by evaluating
CPU wait times to identify workloads that are not meeting
performance objectives
– QoS Management can recommend:
— Adjustments to the size of server pools
r a ble
— Alterations to consumer group mappings
n s fe
— Adjustments to the CPU resources allocated to different
n - tra
database instances within a server pool no a
• Manage memory pressure due to number
) h asof sessions
ฺ or
r e
runaway workloads mฺb uid
– QoS Management restricts oฺnew e
o
c sessions tG
n from being
i k d
e Stusuffering from memory stress
@t that are
established on servers
r te this
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
o T
QoS Management works with Oracle RAC, Oracle Clusterware and Cluster Health Monitor
i c
n to manage database resources to meet service levels, and manage memory pressure
(CHM)
e c
T managed servers.
for
Typically, database services are used to group related work requests and for measuring and
managing database work. For example, a user initiated query against the database might use
a different service from a report generation application. To manage the resources used by a
service, some services may be deployed on several Oracle RAC instances concurrently,
while others may be deployed on only one instance.
In an Oracle RAC database, QoS Management monitors the nodes on which user-defined
database services are offered. Services are created in a specific server pool and the service
runs on all servers in the server pool. If a singleton service is required because the application
cannot effectively scale across multiple RAC servers, the service can be hosted in a server
pool with a maximum size of one.
QoS Management periodically evaluates database server CPU wait times to identify
workloads that are not meeting performance objectives. If needed, QoS Management
provides recommendations for adjusting the size of the server pools or alterations to
Database Resource Manager (DBRM) consumer group mappings. Starting with Oracle
Database release 11.2.0.3, QoS Management also supports moving CPUs between
databases within the same server pool.

The DBRM is an example of a resource allocation mechanism; it can allocate CPU shares
among a collection of resource consumer groups based on a resource plan specified by an
administrator. A resource plan allocates the percentage of opportunities to run on the CPU.
QoS Management does not adjust DBRM plans; it activates a shared multi-level resource
plan and then, when implementing a recommendation, it moves workloads to specific
resource consumer group to meet performance objectives for all the different workloads.
Enterprise database servers can run out of available memory due to too many open sessions
or runaway workloads. Running out of memory can result in failed transactions or, in extreme
cases, a reboot of the server and loss of a valuable resource for your applications. QoS
Management eases memory pressure by temporarily shutting down the services for database
instances on a server suffering from memory stress. This causes new sessions to be directed
to lighter loaded servers. Rerouting new sessions protects the existing workloads and the
availability of the memory-stressed server.
When QoS Management is enabled and managing an Oracle Clusterware server pool, it
le
fe rab
receives a metrics stream from Cluster Health Monitor that provides real-time information
about memory resources for a server, including the amount of available memory, the amount
tra ns
of memory currently in use, and the amount of memory swapped to disk for each server. If
n -
a no
QoS Management determines that a node is under memory stress, the Oracle Clusterware
managed database services are stopped on that node preventing new connections from being
a s
h eฺ
created. After the memory stress is relieved, the services are restarted automatically and the
r )
ฺ Guid
b
listener can send new connections to the server. The memory pressure can be relieved in
m
k o ฺco dent
several ways, for example, by existing sessions closing or by user intervention.
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

QoS Management Policy Sets
QoS Management Policy Set

Maintenance Policy
Weekend Policy
Server
Pools After Hours Policy
Business Hours Policy le
fe rab
a ns
Performance n-tr
no
a
Classes Business
) h asRankingsฺ
b r
ฺ Gu Allocations i d e
o mServert Pool
i k oฺc uden
@ te St
o r te this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
A central
o Tconcept in QoS Management is the policy set. A policy set allows you to specify
i c
ecn
your resources, performance classes (workloads), and a collection of performance policies
T that specify the performance objective for each performance class and sets constraints for
resource availability. QoS Management uses a system-wide policy set that defines
performance objectives based upon the classes of work and the availability of resources.
Specific performance policies can be enabled based upon a calendar schedule, maintenance
windows, events, and so on. Only one performance policy can be in effect at any time.
To maintain the current performance objectives, QoS Management makes resource
reallocation recommendations and predicts their effect. The recommendations can be easily
implemented with a single button click.
A policy set consists of the following:
• The server pools that are being managed by QoS Management.
• Performance classes, which are work requests with similar performance objectives.
• Performance policies, which describe how resources should be allocated to the
performance classes by using performance objectives and server pool directive
overrides. Within a performance policy, performance objectives are ranked based on
business importance which enables QoS Management to focus on specific objectives
when the policy is active.

Server Pools
Sales Sales ERP HR Batch

App Cart App App App
ERP HR Batch
Service Service Service
SALES Service
le
fe rab
trans
n -
Online Server Pool (SP)
Min=3, Max=4,
BackOffice SP
Min=2, Max=2, a
Min=1,
no
Batch SP
Max=1,
Free SP
Min=0,
Importance=30 Importance=20 a
h eฺ s
Importance=10 Imp=0
r )
ฺb Database id
SALES Database APPS
m G u
k o ฺco dent
t
Oracle eiClusterware
S tu
r @
te thi s
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
o
Teik licen
A server pool is a logical division of a cluster. Server pools facilitate workload isolation within
n i co
a cluster while maintaining agility and allowing users to derive other benefits associated with
Te c
consolidation. Administrators can define server pools which are typically associated with
different applications and workloads. An example is illustrated on the slide. QoS Management
can assist in by managing the size of each server pool and also managing the allocation of
resources within a server pool.
When Oracle Grid Infrastructure is first installed, a default server pool, called the Free pool, is
created. All servers are initially placed in this server pool. Specific server pools can then be
created for each of the workloads that needs to be managed. When a new server pool is
created, the servers assigned to that server pool are automatically moved out of the Free pool
and placed into the newly created server pool.
After a server pool is created, a database can be configured to run on the server pool, and
cluster-managed services can be established for applications to connect to the database.
For an Oracle RAC database to take advantage of the flexibility of server pools, the database
must be created using the policy-managed deployment option, which places the database in
one or more server pools.

A key attribute of policy-based management is the allocation of resources to server pools
based on cardinality and importance.
When the cluster starts or when servers are added, all the server pools are filled to their
minimum levels in order of importance. After the minimums are met, server pools continue to
be filled to their maximums in order of importance. If there are any left-over servers, they are
allocated to the Free pool.
If servers leave the cluster for any reason, a server reallocation may take place. If there are
servers in the Free pool and another server pool falls below its maximum value, a free server
is allocated to affected server pool. If there are no free servers, then server reallocation takes
place only if a server pool falls below its minimum level. If that occurs, a server will be
sourced from one of the following locations in the following order:
1. The server pool with the lowest importance that has more than its minimum number of
servers.
2. The server pool with the lowest importance that has at least one server and has lower le
importance than the affected server pool.
fe rab
tr a ns
Using these mechanisms, server pools can maintain an optimal level of resources based on
the current number of servers that are available.
n -
no
Consider the example shown on the slide. If one of the servers in the Online server pool
a
a s
failed, the server currently residing in the Free server pool would automatically move to the
h eฺ
Online server pool. r )
ฺ Guid
b
m
Now, if one of the servers from the BackOffice server pool failed, there would be no servers to
k o ฺco dent
allocate from the Free server pool. In this case, the server currently servicing the Batch server
t ei Stu
pool would be dynamically reallocated to the BackOffice server pool because the failure would
o r te@ this
cause the BackOffice server pool to fall below its minimum and it has a higher importance
than Batch.
ฺ s up use
If one node is later returned to the cluster, it will be allocated to the Batch pool in order to
( t ec se to
satisfy the minimum for that server pool.
e i ko licen
Any additional nodes added to the cluster after this point will be added to the Free pool since
o T
all the other pools are filled to their maximum level.
ic
Tecn

Performance Classes
• A performance class is a group of work requests whose

service level needs to be managed

• Work requests are defined by performance classifiers
containing the database service name and optional
session parameters
• Initial set of performance classifiers are automatically
discovered and created from cluster-managed services
r a ble
• Performance objectives are defined on performance nsfe
classes n - tra
o n
a
) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Performance
o T classes are used to categorize workloads with similar performance
n i c
requirements. A set of classification rules are evaluated against work requests when they
T e c
arrive at the edge of the system. These rules allow value matching against attributes of the
work request; when there is a match between the type of work request and the criteria for
inclusion in a performance class, the work request is classified into that performance class.
This classification of work requests applies the user-defined name, or tag, that identifies the
performance class (PC) to which the work request belongs. All work requests that are
grouped into a particular PC have the same performance objectives. In effect, the tag
connects the work request to the performance objective that applies to it. Tags are carried
along with each work request so that every component of the system can take measurements
and provide data to QoS Management for evaluation against the applicable performance
objectives.
QoS Management supports user-defined combinations of connection parameters called
classifiers to map performance classes to the actual workloads running in the database.

These connection parameters fall into two general classes and can be combined to create
fine-grained Boolean expressions:
• Configuration Parameters: The supported configuration parameters are
SERVICE_NAME and USERNAME. Each classifier in a performance class must include
one or more cluster-managed database services. Additional granularity can be achieved
by identifying the Oracle Database user that is making the connection from either a
client or the middle tier. The advantage of using these classifiers is that they do not
require application code changes to define performance classes.

• Application Parameters: The supported application parameters are MODULE,
ACTION, and PROGRAM. These are optional parameters set by the application as
follows:
- OCI: Use OCI_ATTR_MODULE and OCI_ATTR_ACTION
- ODP.NET: Specify the ModuleName and ActionName properties on the
OracleConnection object le
- JDBC: Set MODULE and ACTION in SYS_CONTEXT
fe rab
a ns
The PROGRAM parameter is set or derived differently for each database driver and platform.
tr
n -
Please consult the appropriate Oracle Database developer's guide for further details and
examples. no
a
a s
h directs
To manage the workload for an application, the application)code
r d e ฺ database
b
connections to a particular service. The service name isฺ specified
m i
u in a classifier, so all work
requests that use that service are tagged as belonging o
ฺc den to t
the G
performance class created for
k o
that application. If you want to provide more
t ei precise
S t ucontrol over the workload generated by
various part of the application, you can
r t e @ his
create additional performance classes and use
classifiers that include MODULE,
p o ACTION,e t or PROGRAM in addition to the SERVICE_NAME
or USERNAME. u
ฺs to u s
t e c
The performance
i k o ( nused
classes
e se in an environment can change over time. A common scenario
Tedividing
is to replace c
a singleliperformance objective with multiple, more-specific performance
i c o
objectives, the work requests into additional performance classes. For example,
n
Tec
application developers can suggest performance classes for QoS Management to use. In
particular, an application developer can define a collection of database classifiers using the
MODULE and ACTION parameters, and then put them in separate performance classes so
each type of work request is managed separately.

Classification and Tagging
• Each session is classified:

– The classification is determined by evaluating session

parameters against the performance class classifiers
– Occurs only when a session is established or when session
parameters change
— This minimizes the overhead associated with classification
• Each work request is tagged: le
– The tag is based on the current session classification fe rab
a ns
n-tr class
– The tag connects the work request with a performance
o
n request to be
— It enables measurements associated with the a
work
has class
recorded against the appropriate performance ฺr) uide
ฺ b
ฺ c om nt G
i k o ude
t e S t
r @
te thi s
up© 2012,uOracle
ฺ s
( t ec se to
Copyright
e i ko licen
i c o TQoS Management, work requests must be classified and tagged.
To enable
n a database session is established, the session parameters are evaluated against the
ecWhen
T performance class classifiers to determine a classification. Work associated with the session
is then tagged based on the session classification until the session ends or the session
parameters change. If the session parameters change, the classification is re-evaluated. Thus
the overhead associated with classification is very small since the classification is only
evaluated when a session is established or when session parameters change.
Tags are permanently assigned to each work request so that all the measurements
associated with the work request can be recorded against the appropriate performance class.
In effect, the tag connects the work request to a performance class and its associated
performance objective.

Performance Policies
• Performance policies are named sets of performance

objectives and server pool overrides to meet business

objectives
– Performance objectives can be ranked according to their
importance
• Only one policy is active at any time
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
To manage
o T various performance objectives, a QoS Management administrator defines one or
i c
ecn
more performance policies. For example, the administrator might define a performance policy
T for normal business hours, another for weekday non-business hours, one for weekend
operations, and another to be used during processing for the quarter-end financial closing.
Note that at any time, only one performance policy is in effect.
A performance policy has a collection of performance objectives in effect; one or more for
each application that is being managed on the system. Some performance objectives are
always more critical to the business than others, while other performance objectives might be
more critical at certain times, and less critical at other times. The ability to define multiple
performance policies inside the policy set provides QoS Management with the flexibility
required to implement different priority schemes when they are required.

Performance Class Ranks
• Performance class ranks assign a relative level of

business criticality to each performance class within a

performance policy:
– Highest
– High
– Medium
– Low le
fe rab
– Lowest
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T
Withinoa performance policy, you can also rank each performance class. This rank assigns a
i c
n level of business criticality to each performance objective. When there are not enough
relative
T e c
resources available to meet all the performance objectives for all performance classes, the
performance objectives for the more critical performance classes must be met at the expense
of the less critical ones. The available rank settings are Highest, High, Medium, Low, or
Lowest. Note that if more than one class is assigned a particular rank (for example, Medium),
classes are then ordered within that ranking alphabetically.

• Performance objectives can be derived from your SLAs.

They specify:
– A business requirement, and
– The performance class to which it applies
• Average response time per database call is currently the
only performance objective type
– Response time is the total time from the time the database
r a ble
receives the request to when the response leaves the server
n s fe
a
n-tr
– Response time does not include network traffic time
no
a
) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
You create
o T a performance objective for each performance class to specify the desired
n i c
performance level for that performance class. A performance objective specifies both a
T e c
business requirement, and the work to which it applies (the performance class). For example,
a performance objective might say that database work requests that use the SALES service
should have an average response time of less than 60 milliseconds.
Each performance policy includes a performance objective for each and every performance
class, unless the performance class is marked measure-only. In this release, QoS supports
only one type of performance objective, average response time.
Response time is based upon database client calls from the point that the database server
receives the request over the network until the request leaves the server. Response time
does not include the time it takes to send the information over the network to or from the
client. The response time for all database client calls in a performance class is averaged and
presented as the average response time.

Performance Satisfaction Metrics
• Different performance objectives can be compared using

the Performance Satisfaction Metric (PSM)

• The PSM quickly shows how the system is coping with the
objective
100%
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco den-100% t
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Different
o Tperformance objectives are used to measure the performance of different workloads.
i c
n Management currently supports only OLTP workloads and uses only the average
QoS
T e c
response time performance objective. When configuring QoS Management, you can have
very different performance objectives for each performance class. For example, one
performance objective may specify that a Checkout call should complete within 1 millisecond,
while another performance objective may specify that a Browse call should complete within 1
second. As more performance objectives are added to a system, it can be difficult to compare
them quickly.
Because of this, it is useful to have a common and consistent numeric measure indicating
how the current workload for a performance class is measuring up against its current
performance objective. This numeric measure is called the Performance Satisfaction Metric.
The Performance Satisfaction Metric is thus a normalized numeric value (between +100%
and -100%) that indicates how well a particular performance objective is being met, and which
allows QoS Management to compare the performance of the system for widely differing
performance objectives.

Server Pool Directive Overrides
Business Hours Policy

Online SP Back Office SP Batch SP
After Hours Policy
le
fe rab
tr ans
Online SP Back Office SP -
Batch SP
n
End Of Quarter Policy
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t eiOffice S tu
Online SP
r t e @ Back
h i s SP Batch SP
u p o se t
c ฺs © t2012,
Copyright
t e
i k o ( ense
Te policyliccan also include a set of server pool directive overrides. A server pool
A performance
o
ic override sets the minimum server count, maximum server count and importance
ecn
directive
T attributes for a server pool when the performance policy is in effect. Server pool directive
overrides serve as constraints on the recommendations proposed by QoS Management
because the server pool directive overrides are honored while the performance policy is
active. For example, QoS Management will never recommend moving a server out of a server
pool if doing so will leave the server pool below its minimum server count value.
Server pool directive overrides can be used to define the normal state of server pools at
different points in time. The slide illustrates an example. Under normal conditions, these
server pool settings would be expected to handle the prevailing workload. If there is a sudden
increase in the workload requests for a performance class, then the associated server pool
might require additional resources beyond what is specified in the performance policy.

Overview of Metrics
• QoS Management uses a standardized set of metrics

• There are two metric types:

– Performance metrics give an overview of where time is spent
in the system
– Resource metrics measure the time that work requests use a
resource or wait for a resource
• Metrics are used to identify bottlenecked resources and to able
determine the best corrective action: s fer
n tisr the a
– For a performance class, the bottlenecked resource n -
no time
resource that contributes the largest averageawait
) h as ฺ
m ฺbr Guide
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
QoS Management uses a standardized set of metrics, which are collected by all the servers in
i c
n system. There are two types of metrics: performance metrics and resource metrics. These
the
T e c
metrics enable direct observation of the use and wait time incurred by work requests in each
performance class, for each resource requested, as it traverses the servers, networks and
storage devices that form the system.
Performance metrics are collected at the entry point to each server in the system. They give
an overview of where time is spent in the system and enables comparisons of wait times
across the system. Data is collected periodically and forwarded to a central point for analysis,
decision-making, and historical storage.
A key performance metric is response time, or the difference between the time a request
comes in and the time a response is sent out. The response time for all database calls in a
Performance Class is averaged and presented as the Average Response Time. Another
important performance metric is the arrival rate of work requests. This provides a measure of
the demand associated with each Performance Class.

Resource metrics exist for the following resources; CPU, Storage I/O, Global Cache, and
Other (database waits). Two resource metrics are provided for each resource:
• Resource usage time - measures how much time is spent using the resource.
• Resource wait time - measures the time spent waiting to get the resource.
QoS Management metrics provide the information needed to systematically identify
performance class bottlenecks in the system. When a performance class is violating its
performance objective, the bottleneck for that performance class is the resource that
contributes the largest average wait time for each work request in that performance class.
le
fe rab
tra ns
n -
a no
a s
h eฺ
r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

QoS Management Architecture
Data Sources QoS Management Enterprise

Server Manager
Oracle
Data Connectors and Manager

RAC JMX Client Management
11.2 JDBC
Interfaces Dashboard
XML
Policy Set
Wizard
Oracle
ble
Performance era
Clusterware Policy Engine
11.2 SRVM
Oracle Business Rules History ns
f
n - tra
Operating a Actions noAlerts &
System a
h eฺ s
Persistencer)
CHM
Manager m ฺ Guid
b
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
QoS Management retrieves metrics data from each database instance running in managed
i c
n pools and correlates the data by performance class every 5 seconds. The data
server
T e c
includes many metrics; for example, call arrival rate and CPU, I/O and Global Cache use, and
wait times. The data is combined with the current topology of the cluster and the health of the
servers in the Policy Engine to determine the overall performance profile of the system with
regard to the current performance objectives established by the active performance policy.
The performance evaluation occurs once a minute and results in a recommendation if there is
a performance class not meeting its objective. The recommendation specifies what resource
is bottlenecked. Specific corrective actions are included, if possible, along with the projected
impact on all performance classes in the system. The slide shows the collection of data from
various data sources by the data connectors component of QoS Management:
• Oracle RAC 11.2 communicates with the data connector using JDBC.
• Oracle Clusterware 11.2 communicates with the data connector using the SRVM
component of Oracle Clusterware.
• The server operating system communicates with the data connector using Cluster
Health Monitor (CHM).
Enterprise Manager displays the information in a variety of ways, including the Management
Dashboard, Policy Set Wizard, Performance History, and Alerts and Actions screens.

QoS Management Recommendations
• If performance objectives are not being met, QoS

Management makes a recommendation

• Each recommendation focuses on improving the highest
ranked performance class exceeding its performance
objective
• Recommendations may include:
le
– Changing consumer group mappings
fe rab
— Reprioritize work within existing resource boundaries
tra ns
– Moving servers between server pools n -
Reprioritize resources between server pools to
no
ameet workload
—
a s
demands
ฺ within )h
iserver pool
ฺ
br de
– Moving CPUs between databases
o m G u
a
Reprioritize CPU resources k o nt server pool
ฺcwithindeexisting
—
boundaries t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T experiences periodic demand surges, then to retain performance levels for
If yourobusiness
i c
n applications you can acquire additional hardware to be available when needed, and sit
your
T e c
idle when not needed. Rather than have extra servers sit idle for most of the time, you might
decide to use those servers to run other application workloads. However, if the servers are
busy running other applications when a demand surge hits, your main business applications
are not able to perform as expected. QoS Management helps to manage such situations.
When you implement a performance policy, QoS Management continuously monitors the
system and manages it using an iterative process. When one or more performance objectives
are not being met, each iteration seeks to improve the performance of a single performance
objective; the highest ranked performance objective that is currently not being met. When all
performance objectives are being met, QoS Management makes no further
recommendations.
The recommendations take the form of moving servers between server pools, changing
consumer group mappings, or moving CPUs between databases within a server pool.
Changing consumer group mappings may involve promoting a specific workload so that it
gets a greater share of resources, or it may involve demoting a competing workload as a way
of making additional resources available to the target performance class. In both cases,
workloads are reprioritized within existing resource boundaries.

Moving servers between server pools is another approach used by QoS Management. This
approach alters the distribution of servers to meet workload demands.
Commencing with Oracle Database release 11.2.0.3, QoS Management can also move CPU
resources between databases within the same server pool. This alters the distribution of CPU
resources between database instances using instance caging and provides additional control
for environments where multiple databases are consolidated within the same Exadata Database
Machine environment.
le
fe rab
tra ns
n -
a no
a s
h eฺ
r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

Implementing Recommendations
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Wheno T Management is working to improve the performance of a particular performance
QoS
i c
n it recommends to add more of the bottleneck resource (such as CPU time) for that
class,
T e c
performance class, or to make the bottleneck resource available more quickly to work
requests in the performance class.
Implementing a recommendation makes the resource less available to other performance
classes. The negative impact on the performance classes from which the resource is taken
may be significantly smaller than the positive impact on the service that is getting better
access, resulting in a net win for the system as a whole. Alternatively, the performance class
being penalized may be less business critical than the one being helped.
When generating recommendations, QoS Management evaluates the impact to the system
performance as a whole. If the improvement for one performance class is rather small, but the
negative impact on another performance class is large, then QoS Management might report
that the performance gain is too small, and not recommended. If there is more than one way
to resolve the bottleneck, then QoS Management advises the best overall recommendation
factoring in variables such as the calculated impact on all the performance classes along with
the predicted disruption and settling time associated with the action. Using Oracle Enterprise
Manager, you can view the current recommendation and the alternative recommendations.

Performance data is sent to Oracle Enterprise Manager for display in the QoS Management
Dashboard and Performance History pages. Alerts are generated to drive notifications that one
or more performance objectives are not being met or that a problem has developed that
prevents one or more server pools from being managed. As a result of these notifications the
administrator can implement the recommendation.
In this release, QoS Management does not implement the recommendations automatically. It
suggests a way of improving performance, which must then be implemented by the
administrator clicking on the Implement button. After implementing a recommendation, the

system is allowed to settle before any new recommendations are made. This is to ensure stable
data is used for further evaluations and also to prevent recommendations that result in
oscillating actions.
le
fe rab
tra ns
n -
a no
a s
h eฺ
r )
ฺ Guid
b
m
k o ฺco dent
t ei Stu
o r te@ this
ฺ s up use
( t ec se to
e i ko licen
o T
ic
Tecn

Quiz
Oracle Database Quality of Service Management helps to meet

performance objectives by reducing resource usage:

a. True
b. False
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o bT
e c n Database of Service Management helps to meet performance objectives by managing
Oracle
T and reducing resource wait times, not resource usage.

Quiz
Oracle Database Quality of Service Management

recommendations can include:

a. Moving servers between server pools
b. Adding spindles to improve I/O performance
c. Changing consumer group mappings
d. Recommending partitioning strategies to ease global
le
cache bottlenecks
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Answer:
i c o a,T c
e c n Database Quality of Service Management can identify I/O performance issues or
Oracle
T global cache bottlenecks, but cannot address them in this release.

Summary

• Describe the purpose of Oracle Database Quality of

Service (QoS) Management
• Describe the benefits of using Oracle Database QoS
Management
• Describe the components of Oracle Database QoS
Management r a ble
• Describe the operation of Oracle Database QoS n s fe
Management n - tra
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

– Configuring Quality of Service Management

– Using Quality of Service Management
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Managing Exadata Database Machine with

Enterpise Manager Cloud Control 12c
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Objectives
After completing this lesson, you should be able to outline the

key Exadata Database Machine management capabilities

included in Enterprise Manager Cloud Control 12c.
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn
Exadata Database Machine Administration Workshop A - 2

Lesson Overview
• This lesson provides an outline of some key Exadata-

specific enhancements in Enterprise Manager Cloud

Control 12c.
• Areas covered in this lesson include:
– Configuring Exadata Database Machine as an Enterprise
Manager target
– Visualizing Exadata Database Machine in Enterprise
r a ble
Manager
n s fe
– Exadata Storage Server management and performance
n - tra
monitoring a no
as ฺ monitoring
– Exadata Database Machine InfiniBand hnetwork
)
r uide
and administration mฺb co ent G
o ฺ
t eik Stud
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T provides an outline of some of the Exadata-specific monitoring and administration
This lesson
o
n i c
enhancements included in Enterprise Manager Cloud Control 12c. The following areas are
T e c
covered:
• Configuring Exadata Database Machine as an Enterprise Manager target.
• Visualizing Exadata Database Machine in Enterprise Manager.
• Exadata Storage Server management and performance monitoring.
• Exadata Database Machine InfiniBand network monitoring and administration.
The intention of this lesson is to provide an outline of some of the key areas which are
enhanced in the new release of Enterprise Manager.

Configuring Exadata Database Machine as an
Enterprise Manager Target
• New target type for Exadata Database Machine
• New wizard-driven guided discovery process:
1. Install Agent 2. Launch 3. Assign 4. Specify 5. Specify

on Compute Auto Monitoring Component Monitoring 6. Review
Node Discovery Agents Credentials Properties
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The new o TExadata Database Machine target type in Enterprise Manager Cloud Control 12c
i c
n the basis for managing Database Machine as an integrated whole rather than as a
provides
T e c
series of related components. This fundamental enhancement to the Enterprise Manager
architecture underpins many of the Database Machine specific monitoring and management
improvements in this release.
Compared to earlier releases, the process of configuring Enterprise to manage Exadata
Database Machine is greatly simplified. The new wizard-driven guided discovery process
makes configuration significantly easier and faster. There are no separate plug-ins to install
and the discovery of Database Machine components is done automatically by the Enterprise
Manager agent.

Visualizing Exadata Database Machine in
Enterprise Manager
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Enterprise
o TManager Cloud Control 12c introduces a new user interface for visualizing Exadata
n i c
Database Machine in its components. It provides an integrated view of the Database Machine
T e c
hardware and software components. The slide shows a selection of screen images.
The left-most image shows a graphical representation of a Full Rack Exadata Database
Machine. Using this view administrators can easily relate severs and switches back to their
physical location in the rack. This is particularly useful when a hardware component raises an
alert and the administrator can easily identify the component with a high degree of certainty.
The images on the right side of the screen show charts relating to different cell and database
server metrics. These server-organized screen layouts are specific to Database Machine,
making it easy for administrators to relate the information on the screen back to specific
Database Machine components.
In addition to the hardware-organized Database Machine views, the are numerous system
and software oriented views which allow administrators to monitor performance and
availability along with resource usage by databases, services and clusters. Software alerts for
databases, clusters and ASM are also presented. A topology-organized view of Database
Machine clusters and database is also provided while the built-in configuration management
repository tracks the version information associated with all the Database Machine
components.

Exadata Storage Server Monitoring
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Enterprise
o TManager Cloud Control 12c includes enhanced group-based monitoring and
n i c
administration capabilities. By default, all the cells in a Database Machine are grouped into a
T e c
cell group. As shown in the screen image, consolidated monitoring capabilities are provided
for the cell group. Group-level administration operations are also provided so that actions can
be applied consistently across all the cells in the group.

Exadata Storage Server Administration
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
Enterprise
o TManager Cloud Control 12c includes the ability to perform cell configuration and
n i c
administration tasks. The slide shows an example of the interface that can be used to enable,
T e c
disable and configure I/O Resource Manager (IORM). Enterprise Manager Cloud Control 12c
also includes a range of other cell administration functions such as the ability to start and stop
cell services, verify cell connectivity and configure SSH access.

Exadata Storage Server Performance Monitoring
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The screen
o T image on the slide show an example of the Exadata Storage Server performance
n i c
monitoring capabilities in Enterprise Manager Cloud Control 12c. The information presented
T e c
by Enterprise Manager provides a composite view of the performance indicators for a single
cell or cell group. When users point at different chart elements, additional information is
displayed.

Exadata Storage Server Health Monitoring
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
The screen
o T image on the slide show an example of the Exadata Storage Server health
n i c
monitoring capabilities in Enterprise Manager Cloud Control 12c. The information presented
T e c
by Enterprise Manager provides a composite view of the health indicators for a cell group.
A slider enables the targeted examination of a specific time period, More detailed information
for the selected time period can also be examined. The detailed information can relate to
issues such as response degradation, I/O load imbalances, ASM related problems, cell
software or hardware failures, cell configuration issues and network related failures.

InfiniBand Network Monitoring
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T Manager Cloud Control 12c, the Exadata Database Machine InfiniBand network
In Enterprise
o
i c
ecn
and switches are native targets which are automatically discovered during the guided
T discovery process. Real time and historical performance metrics relating to the InfiniBand
network are available for administrators to analyze. Also, Enterprise Manager will highlight
alert conditions which are generated within the InfiniBand switches or which relate to
thresholds maintained inside Enterprise Manager. Configuration metrics are used to detect
and notify administrators about network changes and situations that violate best-practice
recommendations.

InfiniBand Network Administration
le
fe rab
tra ns
n -
no a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
T Database Machine InfiniBand network administration function are also available
New Exadata
o
inn i c
Enterprise Manager Cloud Control 12c. The screen images in the slide shows an example
e c
T some of the InfiniBand network administration screens. The available administration
of
actions include:
• Enable an InfiniBand port.
• Disable an InfiniBand port.
• Clear performance counters.
• Clear error counters.
• Switch on an indicator LED.
• Switch off an indicator LED.

Summary
In this lesson you should have learned how to outline the key
Exadata Database Machine management capabilities included

in Enterprise Manager Cloud Control 12c.
le
fe rab
tra ns
n -
no
a
a
h eฺ s
ฺ Guid
b r )
m
k o ฺco dent
t ei Stu
o r te@ this
up© 2012,uOracle
( t ec se to
Copyright
e i ko licen
o T
ic
Tecn

Tecnico Teiko (tecฺsuporte@teikoฺcomฺbr) has a non-transferable license to use this Student Guideฺ

Uploaded by

Copyright:

Available Formats

Tecnico Teiko (tecฺsuporte@teikoฺcomฺbr) has a non-transferable license to use this Student Guideฺ

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Tecnico Teiko (tecฺsuporte@teikoฺcomฺbr) has a non-transferable license to use this Student Guideฺ

Uploaded by

Copyright:

Available Formats

Unauthorized reproduction or distribution prohibitedฺ Copyright© 2013, Oracle and/or its affiliatesฺ

Peter Fusek Disclaimer

This document contains proprietary information and is protected by

share, download, upload, copy, print, display, perform, reproduce,

Eric Siglin Varun Malhotra

Harald van Breederode Vijay Sridharan

X2-2 Full Rack Leaf Switch Topology 3-8

5 Exadata Database Machine Initial Configuration

dcli Examples 6-6

7 I/O Resource Management

Complete Example 7-22

8 Recommendations for Optimizing Database Performance

10 Consolidation Options and Recommendations

12 Bulk Data Loading using Oracle DBFS

Bulk Data Loading using Oracle DBFS Overview 12-3

14 Configuring Enterprise Manager Grid Control 11g to Monitor Exadata Database

Deploying the System Monitoring Plug-in for Oracle Exadata Power

Monitoring Oracle Grid Infrastructure 16-6

17 Monitoring the InfiniBand Network le

19 Other Useful Monitoring Tools

20 Backup and Recovery

22 Patching Exadata Database Machine

Test System Recommendations 22-11

23 Exadata Database Machine Automated Support Ecosystem

A Managing Exadata Database Machine with Enterpise Manager Cloud

Bulk Data Loading using Oracle DBFS

After completing this lesson, you should be able to:

• Describe the process of using Oracle DBFS for bulk data

Exadata Database Machine Administration Workshop 12 - 2

Source System Database Machine

Staging Database CREATE ... AS SELECT ...

k o ฺco dent data

Exadata Database Machine Administration Workshop 12 - 3

• Data files can be in any format supported by the external

Exadata Database Machine Administration Workshop 12 - 4

• Data files should always be staged in DBFS

– Not database server internal drives

Exadata Database Machine Administration Workshop 12 - 5

DBFS should be configured in a separate staging database

• Use DBCA to create the staging database

Exadata Database Machine Administration Workshop 12 - 6

• Additional database server operating system configuration:

Exadata Database Machine Administration Workshop 12 - 7

• Creating the DBFS store

Exadata Database Machine Administration Workshop 12 - 8

Exadata Database Machine Administration Workshop 12 - 9

• Prerequisites for data file access using external tables

– Create an Oracle directory object that references the DBFS

Exadata Database Machine Administration Workshop 12 - 10

• Recommended approach uses external tables

– Parallel direct path loading for high-performance

Exadata Database Machine Administration Workshop 12 - 11

• Parallel loading using external tables

– A CREATE TABLE ... AS SELECT statement

• Direct path loading using external tables le

Exadata Database Machine Administration Workshop 12 - 12

To facilitate parallel loading from typical flat files external table

definitions must reference numerous smaller data files rather

Exadata Database Machine Administration Workshop 12 - 13