h9538 Vfcache Oracle Vmax WP

White Paper
EMC Solutions Group

Abstract
This white paper describes the testing of EMC VFCache with EMC Symmetrix
VMAX to accelerate online transaction processing (OLTP) performance in an
Oracle Database 11g R2 environment. Though the testing was performed on
Symmetrix VMAX 40K, the capabilities and benefits of VFCache are applicable
to Symmetrix VMAX 10K.
November 2012
EMC VFCACHE ACCELERATES ORACLE
EMC VFCache, EMC Symmetrix VMAX and VMAX 10K,
Oracle Database 11g
VFCache dramatically improves Oracle performance
Symmetrix VMAX and VMAX 10K protects data

EMC VFCache Accelerates Oracle
EMC VFCache, EMC Symmetrix VMAX and VMAX 10K, Oracle Database 11g

2

Copyright 2012 EMC Corporation. All Rights Reserved.
EMC believes the information in this publication is accurate as of its
publication date. The information is subject to change without notice.
The information in this publication is provided as is. EMC Corporation makes
no representations or warranties of any kind with respect to the information in
this publication, and specifically disclaims implied warranties of
merchantability or fitness for a particular purpose.
Use, copying, and distribution of any EMC software described in this
publication requires an applicable software license.
For the most up-to-date listing of EMC product names, see EMC Corporation
Trademarks on EMC.com.
All trademarks used herein are the property of their respective owners.
Part Number: H9538.1

3

Table of contents
Executive Summary ............................................................................................................................ 4
Business case .................................................................................................................................. 4
The storage performance challenge .................................................................................................. 4
The VFCache solution ....................................................................................................................... 5
Support for Oracle Real Application Cluster (RAC) ............................................................................ 7
Stale data ........................................................................................................................................ 7
Solution architecture .......................................................................................................................... 8
Introduction ..................................................................................................................................... 8
Physical architecture ........................................................................................................................ 8
EMC Symmetrix VMAX ...................................................................................................................... 8
VFCache technology with Oracle Database: testing and validation ...................................................... 9
Introduction ..................................................................................................................................... 9
Note ................................................................................................................................................. 9
Workload profile .............................................................................................................................. 9
Recommended configuration ........................................................................................................... 9
Performance characteristics ............................................................................................................. 9
Conclusion ....................................................................................................................................... 12
References ....................................................................................................................................... 13
White papers ................................................................................................................................. 13
Product documentation .................................................................................................................. 13


4
Executive Summary

In an increasingly competitive environment, businesses are being driven to optimize
business processes and to improve service, while lowering IT costs. Meeting these
new requirements has become critical to the financial success of many companies.
Consequently, operational and revenue-generating applications are experiencing
dramatic demands on performance, driven by:
Growth in the numbers of active users
Time-critical applications and escalating service-level agreements
Increased complexity of business processes and new analytic workloads
Multiple databases with high concurrent access
Businesses need to consider new approaches to performance challenges in order to
meet these demands cost-effectively and without sacrificing data protection.

The latest servers with multi-core processors represent a potential performance
bottleneck for the storage subsystem. As processing capacity and heavier workloads
are added, the storage system is challenged to keep pace with the growing I/O
demands. While CPU performance improves 100-fold every decade, magnetic disk
remains relatively flat, as shown in Figure 1.

Figure 1. CPU performance versus disk drives
In a traditional architecture, as shown in Figure 2:
Reads and writes are serviced by the storage array
Performance varies depending on the back-end arrays media, workload, and
network

Business case
The storage
performance
challenge

5

Figure 2. Traditional architecture
What if you could double your application performance by decreasing latency and
increasing IOPS and transactions per minute?

EMC

VFCache is a server Flash caching solution that uses intelligent caching
software and PCIe Flash technology to reduce latency and increase throughput, which
dramatically improves application performance.
VFCache brings Flash memory performance into the server. The caching optimization
within VFCache automatically adapts to changing workloads by determining which
data is most frequently referenced and promoting it to the server Flash cache.
VFCache enhances both virtualized and bare-metal applications so you can smoothly
migrate your data center to a private cloud at a pace that makes sense for your
business.
When using VFCache, the hottest data automatically resides on the PCIe card in the
server, providing significantly faster access.
VFCache protects data by using a write-through algorithm, which means that writes
persist to the back-end storage array. EMC
trusted networked storage, such as EMC

Symmetrix VMAX and EMC VNX family of storage arrays, protect data with
advanced data services, which include high availability, data integrity, reliability, and
disaster recovery, as shown in Figure 3.
The VFCache
solution


6

Figure 3. VFCache accelerates I/O performance within the server
VFCache is transparent to applications. When VFCache is present, applications do not
have to be rewritten, recertified, or retested.
As shown in Figure 4, VFCache dramatically accelerates the performance of read-
intensive applications:
Reads are serviced by VFCache for performance
Writes are passed through to the storage array for protection

Figure 4. Advanced architecture with EMC VFCache

7

While this white paper focuses on Oracle Database 11g, the VFCache architecture can
directly enhance the performance of many other applications and indirectly enhance
write-intensive applications that now have greater access to SAN resources.
VFCache software caches the most frequently used data on the server-based PCIe
card, thereby putting the data closer to the application. By operating in the server on
the PCIe bus, VFCache reduces the need to access data across the network from the
storage array. This decreases response time and increases performance.
VFCache provides better performance for the applications that involve heavy read I/O.
In one of the tests, VFCache demonstrated read throughput improvements of 210
percent and reductions in read latency by as much as 60 percent. VFCache works with
applications as diverse as databases, analytics, enterprise application servers, email,
and web servers to give them the performance boost they need.
VFCache offloads much of the read I/O traffic from the storage array, which allows it
to allocate greater processing power to other applications. While one application is
accelerated with VFCache, the arrays performance for other applications is
maintained or even slightly enhanced. As VFCache is installed on more servers in the
environment, the result is a highly scalable I/O processing model. The environment
as a whole, including the servers and the storage system, is capable of processing
increasingly more IOPS.
For more information about EMC VFCache, see Introduction to EMC VFCache.

The current release of VFCache software is not supported in shared disk
environments and active/active clusters, therefore, it does not support Oracle Real
Application Clusters (RAC). The VFCache is not coherent between servers in the
current release, which can lead to possible database corruption if used in Oracle RAC
environments.
This issue does not occur with Oracle non-RAC databases, and all the testing and
results discussed in this white paper have been done on an Oracle non-RAC
database.

Even with Oracle non-RAC databases, you have to be cautious about two possible
scenarios that can leave stale data in the VFCache:
Using storage array snapshots: A LUN that is restored from an earlier snapshot
without the knowledge of VFCache software on the server can potentially leave
stale blocks in VFCache.
Migrating between servers: If the database instance was migrated to a different
server for operational reasons and relocated back to the original server without
the knowledge of VFCache software, this can also result in stale blocks in
VFCache.
To avoid either of these situations, you should purge the cache prior to restoring the
LUNs from snapshot or moving the database between servers. There is no need to
purge the entire cache; you can choose to purge only the LUNs that are affected by
the operation.

Support for Oracle
Real Application
Cluster (RAC)
Stale data


8
Solution architecture

This section provides an overview of the physical architecture of this solution.

Figure 5 shows the physical architecture for this solution.

Figure 5. EMC VFCache with Oracle Database and EMC Symmetrix VMAX
The solution architecture consists of an Oracle Database, Cisco server, and a
Symmetrix VMAX storage array. The solution can be configured with a number of
storage arrays, such as Symmetrix VMAX 10K or VNX. In this solution, we used
Symmetrix VMAX 40K. The server used for this test configuration was Cisco UCS
C-460 rack mount with four Xeon processors. The server configuration included the
VFCache server-based Flash cache and two 8 Gb/s connections to the SAN switches.

EMC Symmetrix VMAX storage arrays provide high performance and scalability for
demanding enterprise storage environments. Built on the strategy of simple,
intelligent, modular storage, the VMAX incorporates a highly scalable Virtual Matrix
Architecture that enables it to grow seamlessly and cost-effectively from an entry-
level configuration into the worlds largest storage system. The VMAX supports Flash
drives, SAS drives, Fibre Channel (FC) drives, and SATA drives within a single array, as
well as an extensive range of RAID types.
Introduction
Physical
architecture
EMC Symmetrix
VMAX

9

VFCache technology with Oracle Database: testing and validation

EMC tested the capabilities of VFCache to accelerate online transaction processing
(OLTP) performance in an Oracle Database 11g R2 non-RAC environment. This section
presents the results of the testing.
Benchmark results are highly dependent upon workload, specific application
requirements, and system design and implementation. Relative system performance
will vary because of these and other factors. Therefore, this workload should not be
used as a substitute for a specific customer application benchmark when critical
capacity planning and/or product evaluation decisions are contemplated.

Our testing employed an OLTP workload, with a 1.2 TB database and a 70-30 percent
read/write mix. We took a performance baseline to validate the performance
characteristics of the environment. We enabled VFCache and then took
measurements of transactions per minute, application-level transaction latency, and
Oracle I/O wait times to demonstrate the performance contribution of VFCache to the
application.
This 1.2 TB database had approximately 250 GB of hot data.

VFCache is enabled on all of the data LUNs. However, VFCache was not enabled on
the log LUNs because they are dominated by writes.
No specific tuning was required for Oracle Database.
With this configuration, VFCache uses its caching algorithms to automatically
maintain a copy of the hottest data for immediate access.

Figure 6 compares the overall system throughput (transactions per minute) of the
baseline and VFCache-enabled environments. Tests were run in various
configurations that had different complements of storage and application
configurations. The availability of the hot data in the servers can provide VFCache
results in the 40 to 210 percent improvement range, as shown in Figure 6.
Introduction
Note
Workload profile
Recommended
configuration
Performance
characteristics


10

Figure 6. Relative OLTP transactions per minute (TPM) improvement
The throughput performance improvement is a result of a dramatic reduction in read
latency, as shown in Figure 7.

Figure 7. Relative application-level latency improvement
VFCache copies and stores the hottest read data on the PCIe card within the server.
This data can be repeatedly read by the application without sending the I/O request
to the back-end storage. In a workload with a 70-30 percent read/write mix, this
resulted in an overall average transaction latency decrease of 60 percent.
The effect of servicing read requests from VFCache can be seen in the distribution of
wait times, as shown in Figure 8. In the baseline results, approximately 60 percent of
requests are serviced within 1 millisecond (ms).

11

By servicing hot data read requests from VFCache, we observed a 50 percent increase
in the number of requests that are serviced within 1 ms or less.

Figure 8. VFCache impact on Oracle I/O wait events


12
Conclusion
Our testing in an Oracle OLTP workload compared a system equipped with VFCache
against a baseline configuration without VFCache. The VFCache-equipped server
demonstrated the following performance advantages:
System throughput, measured in transactions per minute, was up to 3.1 times
the throughput of the baseline, with the configuration and test cases that were
used.
Overall transaction latency was up to 60 percent less than the baseline.
Fifty percent more I/O requests were serviced within 1 ms or less.

13

References

For additional information, see the white papers listed below:
Introduction to EMC VFCache
EMC VFCache Accelerates Virtualized Oracle - EMC VFCache, EMC Symmetrix
VMAX and VMAX 10K, VMware vSphere, Oracle Database 11g
EMC VFCache Accelerates Microsoft SQL Server - EMC VFCache, EMC VNX,
Microsoft SQL Server 2008
EMC VFCache Accelerates Oracle - EMC VFCache, EMC VNX, Oracle
Database 11g
For additional information, see the product documents listed below:
EMC VFCache - Data Sheet
EMC Symmetrix VMAX Data Storage System - Specification Sheet

White papers
Product
documentation

h9538 Vfcache Oracle Vmax WP

Uploaded by

Copyright:

Available Formats

h9538 Vfcache Oracle Vmax WP

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

h9538 Vfcache Oracle Vmax WP

Uploaded by

Copyright:

Available Formats

White Paper

EMC Solutions Group

trusted networked storage, such as EMC

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.