h8930 Vplex Metro Oracle Rac WP

White Paper
Oracle Extended RAC with

EMC® VPLEX™ Metro
Best Practices Planning
Abstract
This white paper describes EMC® VPLEX™ features and functionality
relevant to Oracle Real Application Cluster (RAC) and Database. The best
practices for configuring an Oracle Extended RAC environment to optimally
leverage EMC VPLEX are also presented.
September 2011
Copyright © 2011 EMC Corporation. All Rights Reserved.
EMC believes the information in this publication is accurate 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.
VMware is a registered trademark of VMware, Inc. All other

trademarks used herein are the property of their respective
owners.
Part Number h8930
Oracle Extended RAC with EMC VPLEX Metro 2

Table of Contents
Executive summary.................................................................................................. 4
Audience ............................................................................................................................ 5
Introduction ............................................................................................................ 5
Products and features overview ............................................................................... 6
VPLEX ................................................................................................................................. 6
VPLEX product offerings ................................................................................................. 7
VPLEX architecture highlights ......................................................................................... 9
VPLEX consistency groups, detach rules and Witness ................................................... 11
Oracle Real Application Clusters ....................................................................................... 13
Oracle Automatic Storage Management........................................................................ 13
Oracle RAC on Extended Distance Clusters ................................................................... 13
Symmetrix VMAX .............................................................................................................. 14
Symmetrix VMAX TimeFinder product family ................................................................. 14
Virtual Provisioning ...................................................................................................... 15
Implementing VPLEX Metro with Oracle Extended RAC ............................................ 15
Oracle Extended RAC deployment considerations in a VPLEX Metro environment ............. 15
Oracle Extended RAC and VPLEX Metro deployment guidelines .................................... 16
Oracle Extended RAC and VPLEX Metro protection from unplanned downtime .............. 17
VPLEX Metro with Oracle Extended RAC lab configurations and tests ....................... 22
Lab configuration and setup ............................................................................................. 22
Physical environment ................................................................................................... 22
Storage setup and device allocation planning .............................................................. 25
VPLEX Metro setup ....................................................................................................... 28
Host and Oracle setup .................................................................................................. 33
OLTP database workload tests.......................................................................................... 37
Failure conditions tests .................................................................................................... 38
Conclusion ............................................................................................................ 39
References ............................................................................................................ 39

Executive summary
EMC® VPLEX™ is an enterprise-class storage federation technology that aggregates
and manages pools of Fibre Channel (FC) attached storage within and across data
centers. VPLEX resides between the servers and FC storage and presents local and
distributed volumes to hosts. VPLEX storage aggregation allows online storage
migrations and upgrades without any changes to host LUNs. VPLEX AccessAnywhere
clustering technology allows read/write access to distributed volumes across
distance where the volumes have the exact same SCSI LUN identity. This technology
allows hypervisors to migrate virtual machines (VMs) across distance and
clusterwares like Oracle Real Application Cluster (RAC) to provide high availability
across distance. The VPLEX product line includes VPLEX Local (single site SAN
federation), VPLEX Metro supporting synchronous distributed volumes with round-trip
latency up to 5 ms, and VPLEX Geo support asynchronous distributed volumes with
round-trip time up to 50 ms. This paper highlights VPLEX Metro working in
conjunction with Oracle Extended RAC to provide improved application availability,
resiliency to failures, and concurrent utilization of hardware resources at both data
centers.
Oracle RACs support the transparent deployment of a single database across a cluster
of servers, providing fault tolerance, high availability and scalability. Oracle Extended
RAC is an architecture where servers in the cluster reside in locations that are
physically separate. Oracle Extended RAC provides a way to scale out performance,
utilize storage and server resources at multiple sites, and provide an increased
resiliency to failure scenarios or maintenance operations without application
downtime.
Main benefits of VPLEX Metro and Oracle Extended RAC:
Continuous database availability through network, server, storage and site
failures
Database sessions automatic failover using Oracle Transparent Application
Failover (TAF)
Scale out architecture and full read-write access to the same database at both
sites (no idle hardware at standby site)
Simplification of deployment by using VPLEX:
o Hosts need only to connect to local VPLEX cluster
o No need to deploy Oracle voting disk and clusterware on a third site
o No host CPU cycles consumed on mirroring. IO sent only once from host
o Ability to create consistency groups that protect multiple databases or
application files as a unit

o VPLEX volumes do not require an application downtime or LUN ID changes
during storage hardware refresh and migrations
o Easy Disaster Recovery testing and validation by the fact that both sites
can actively participate in the workload
The EMC Symmetrix® VMAX™ was used in this paper as the storage array behind
VPLEX clusters. Symmetrix VMAX series provides an extensive offering of industry-
leading features and functionality for the next era of high-availability virtual data
centers and mission-critical applications. With advanced levels of data protection and
replication, the Symmetrix VMAX system is at the forefront of enterprise storage area
network (SAN) technology. Additionally, the Symmetrix VMAX array with FAST VP
technology transparently optimizes storage tiers, improving performance and saving
cost without disruption to host applications.
IT organizations are looking for ways to extend their database environments across
data centers to reduce or avoid planned and unplanned downtime associated with
hardware failures, disasters or even normal data center operations such as hardware
refresh and migrations. With EMC VPLEX, such organizations are able to achieve new
levels of flexibility in changes to their storage infrastructure, disaster resilience, and
improved collaboration and scale out architecture. Thus EMC VPLEX system is a
natural fit for this next generation of environments together with Oracle RAC
technology. The capabilities of EMC VPLEX to federate storage systems and present
globally unique devices not bound by physical data centers boundaries work hand in
hand with the inherent capabilities of Oracle RAC to provide high availability and
scalable database access. Oracle RAC together with EMC VPLEX and Symmetrix VMAX
is an ideal choice for deploying a highly reliable cloud computing environment,
reducing IT costs while also increasing infrastructure efficiency.
Audience
This white paper is intended for Oracle database administrators, storage
administrators, and IT architects responsible for architecting, creating, managing, and
using IT environments that focus on high availability with Oracle databases, VPLEX
technologies, and Symmetrix VMAX storage. The white paper assumes readers are
somewhat familiar with Oracle RAC and Oracle database technology, and EMC VPLEX
and the Symmetrix storage array.
Introduction
Oracle Extended RAC provides a way to scale out performance, utilize storage and
server resources at multiple sites, and provide an increased resiliency to failure
scenarios or maintenance operations without application downtime - allowing
organizations to eliminate database downtime, and continue business processing
uninterrupted, even in the case of complete site failures.
It should be noted that while Oracle Stretched RAC provides high availability across a
single database, it is still a good practice to deploy a Disaster Recovery (DR) solution

over longer distances with technologies such as RecoverPoint SRDF®, Oracle Data
Guard, and so on. Such remote replicas can help in case of database failure (such as
erroneous file or LUN deletion, block corruptions, and so on). In addition, it is a good
practice to deploy backup strategy to tape or VTL, potentially using clone/snapshot
technology to offload the backup process from production.
This white paper introduces readers to the EMC VPLEX family, VPLEX Metro cluster
architecture, and features and functionality that are relevant to Oracle Extended RAC
deployment. The paper discusses the improved resiliency of the solutions against
different failure conditions. It also provides the steps of provisioning VPLEX and
Symmetrix VMAX storage for the Oracle Extended RAC platform, and running OLTP
workload in a 4-node Oracle Extended RAC with VPLEX Metro and Symmetrix VMAX.
Products and features overview

VPLEX
EMC VPLEX is a storage virtualization solution for both EMC and non-EMC storage, as
shown in Figure 1. The storage behind VPLEX can be heterogeneous, supporting both
EMC storage and common arrays from other storage vendors, such as NetApp, HDS,
HP, and IBM.
Figure 1. EMC VPLEX Local to federate heterogeneous storage

VPLEX can be extended across geographically dispersed data centers to provide
simultaneous access to storage devices through the creation of VPLEX
Distributed Virtual Volumes. VPLEX technology provides non-disruptive,
heterogeneous data movement and volume management functionality.
Because of these capabilities, VPLEX delivers unique and differentiated value to
address three distinct requirements:

The ability to dynamically move applications and data across different compute
and storage infrastructures, be they within or across data centers.
The ability to create high-availability storage and compute infrastructure
geographically dispersed with unmatched resiliency.
The ability to provide efficient real-time data collaboration over distance.
VPLEX product offerings
EMC offers VPLEX in three configurations to address customer needs for high-
availability and data mobility as seen in Figure 2:
• VPLEX Local
• VPLEX Metro
• VPLEX Geo
Figure 2. VPLEX topologies
VPLEX Local
VPLEX Local provides seamless, non-disruptive data mobility and ability to manage
multiple heterogeneous arrays from a single interface within a data center.
VPLEX Local allows increased availability, simplified management, and improved

utilization across multiple arrays.
VPLEX Metro with AccessAnywhere
VPLEX Metro with AccessAnywhere enables active-active, block level access to data
between two sites within synchronous distances up to 5 ms round-trip time (RTT).
Following are two examples of using VPLEX Metro and Oracle for data mobility and
high-availability.
Application and Data Mobility —By itself, the hypervisor has the ability to move
VMs without application downtime between physical servers. When combined
with server virtualization, VPLEX distributed volumes allow users to transparently

move and relocate VMs and their corresponding applications and data over
distance. This provides a unique capability allowing users to relocate, share, and
balance infrastructure resources between sites. An Oracle VM Live Migration
mobility example with VPLEX Metro is covered in the white paper: Oracle VM
Server for x86 Live Migration with EMC VPLEX and Symmetrix VMAX, as can be
seen in Figure 3.
Figure 3. Oracle VM Live Migration with VPLEX Metro
High Availability Infrastructure — Reduces recovery time objective (RTO). High

Availability provides near continuous uptime for critical applications, and
automates the restart of an application once a failure has occurred, with as little
human intervention as possible. An example of Oracle Extended RAC with VPLEX
Metro is shown in Figure 4. This solution is the focus of this white paper.
Figure 4. Oracle Extended RAC with VPLEX Metro

VPLEX Geo with AccessAnywhere
VPLEX Geo with AccessAnywhere enables active-active, block level access to data
between two sites within asynchronous distances. VPLEX Geo enables better cost-
effective use of resources and power. VPLEX Geo provides the same distributed
device flexibility as Metro but extends the distance up to and within 50 ms RTT. As
with any asynchronous transport media, bandwidth is important to consider for
optimal behavior as well as application access method to distributed devices with
long distances. An example of VPLEX Geo use case is when application 1 addressing
VPLEX distributed volumes in consistency group 1 and performs read/write only at
site 1. Application 2 addressing VPLEX distributed volumes in consistency group 2
and performs read/write only at site 2. VPLEX Geo maintains write IO fidelity for each
application at the remote site, although slightly behind in time. In this example both
applications can fail over and back independently, over long distances.
VPLEX architecture highlights1
The VPLEX family uses a unique clustering architecture to help customers break the
boundaries of the data center and allow servers at multiple data centers to have
read/write access to shared block storage devices. VPLEX Local includes a single
cluster while VPLEX Metro and Geo each include two. A VPLEX cluster consists of one,
two, or four engines as seen in Table 1. Each VPLEX engine provides SAN/WAN
connectivity, cache and processing power with two redundant directors.
Table 1. VPLEX hardware components
Feature Description
VPLEX Cluster Contains one, two, or four engines
VPLEX Engine Contains two directors, management modules, power

supplies, battery power and fans
Contains some of the I/O modules, SSD, CPU and RAM
VPLEX Director
VPLEX Local uses write-through caching and lets writes pass directly and get
acknowledged first at the storage behind the VPLEX volumes before acknowledging
them back to the host. With EMC storage such as Symmetrix and VNX™, where writes
only need to register with the storage persistent cache, application write response
time is optimal. VPLEX Metro also uses write-through cache but will acknowledge the
writes to the application only once they were registered with both local and remote
storage. On the other hand, writes to VPLEX Geo are cached at the VPLEX layer and
sent to the remote cluster in deltas that preserve write-order fidelity. In all three
VPLEX deployments – Local, Metro, and Geo, reads can benefit from the VPLEX cache,
and in VPLEX Metro and Geo, read-hits are served from the local VPLEX cluster cache.
1
The details in this section are based on VPLEX release 5.0.1 and may be different in other releases. The VPLEX product guide
provides exact version details.

VPLEX logical storage structure
VPLEX encapsulates traditional physical storage array devices and applies three
layers of logical abstraction to the storage volumes as seen in Figure 5. Extents are
the mechanism VPLEX uses to divide storage volumes. Extents may be all or part of
the underlying storage volume. EMC VPLEX aggregates extents and it can apply RAID
protection in the device layer. Devices are constructed using one or more extents and
can be combined into more complex RAID schemes and device structures as desired.
At the top layer of the VPLEX storage structures are virtual volumes. Virtual volumes
are created from devices and inherit the size of the underlying device. Virtual volumes
are the elements VPLEX exposes to hosts using its front-end (FE) ports. Access to
virtual volumes is controlled using storage views, which are comparable to Auto-
provisioning Groups on EMC Symmetrix or to storage groups on EMC CLARiiON®. They
act as logical containers determining host initiator access to VPLEX FE ports and
virtual volumes.
Figure 5. EMC VPLEX logical storage structures

VPLEX encapsulates storage area network (SAN) volumes by identifying storage
devices with their WWNs, packaging them into sets of VPLEX virtual volumes with
user-defined configuration and protection levels, and then presenting the virtual
volumes to hosts. VPLEX has the ability to encapsulate/de-encapsulate existing
storage devices (preserving their data), as well as the ability to divide and aggregate
existing storage volumes for use in virtual volumes. Furthermore, virtual storage
presented to hosts through VPLEX encapsulation can be non-disruptively moved
within and between back-end storage arrays. It is recommended that when VPLEX
encapsulates devices of EMC storage arrays such as VMAX and VNX the RAID
protection is done at the storage array. This will keep the mapping between storage
and VPLEX simple and will allow the use of storage features such as creating snaps,
clones, and additional DR.

VPLEX consistency groups, detach rules and Witness
Consistency groups
Starting with GeoSynchrony 5.0 for VPLEX Metro and GeoSynchrony 5.0.1 for VPLEX
Geo, consistency groups are used to organize virtual volumes. Consistency groups
aggregate volumes together to provide a common set of properties to the entire
group. In addition, you can move a consistency group from one cluster to another if
required. Consistency groups are particularly important for databases and
applications. All database LUNs (for example, Oracle data, control and log files)
require preserving write-order fidelity to maintain data integrity, and therefore should
always be placed in a single consistency group together. Often multiple databases
have transaction dependencies, such as when database links are used to connect
databases, or when the application issues transactions to multiple databases and
expect them to be consistent with each other. In these cases the consistency group
should include all LUNs that require preserving IO dependency (write-order fidelity).
There are two types of consistency groups:
Synchronous consistency groups — Provide a convenient way to apply the same

detach rules and other properties to a group of volumes in a VPLEX Local or VPLEX
Metro configuration, simplifying system configuration and administration on large
systems. Volumes in a synchronous group have global or local visibility. Synchronous
consistency groups use write-through caching (known as synchronous cache mode in
the VPLEX user interface) and with VPLEX Metro are supported on clusters separated
by up to 5 ms of latency. This means that VPLEX Metro sends writes to the back-end
storage volumes, and acknowledges the write to the application as soon as the back-
end storage volumes in both clusters acknowledge the write.
Asynchronous consistency groups — Used for distributed volumes in a VPLEX Geo,

separated by up to 50 ms of latency. All volumes in an asynchronous consistency
group share the same detach rules and cache mode, and behave the same way in the
event of an inter-cluster link failure. Only distributed volumes can be included in an
asynchronous consistency group. Asynchronous consistency groups use write-back
caching (known as asynchronous cache mode in the VPLEX user interface). This
means the VPLEX director caches each write and then protects the write on a second
director in the same cluster. Writes are acknowledged to the host once the write to
both directors is complete. These writes are then grouped into deltas. Deltas are
compared and reconciled between both clusters before the data is committed to
back-end storage. Writes to the virtual volumes in an asynchronous consistency
group are ordered such that all the writes in a given delta are written before writes
from the next delta (to preserve write order fidelity). Asynchronous cache mode can
give better performance, but there is a higher risk that some data will be lost during a
VPLEX cluster failure or inter-cluster link failure due to the Async nature of the
replication, or if writes are performed at both local and remote locations.

Detach rules
Detach rules are predefined rules that determine consistency group I/O processing
semantics whenever connectivity with remote cluster is lost (for example network
partitioning or remote cluster failure). In these situations, until communication is
restored, most workloads require specific sets of virtual volumes to resume I/Os on
one cluster and suspend I/Os on the other.
In a VPLEX Metro configuration detach rules can depict a static preferred cluster by
setting2: winner:cluster-1, winner:cluster-2 or No Automatic Winner (where the last
one specifies no preferred cluster). When the system is deployed without VPLEX
Witness (discussed in the next section), I/Os to consistency group devices proceeds
on the preferred cluster and are suspended on the non-preferred cluster.
Asynchronous consistency groups detach rules can be Active Cluster Wins or No

Automatic Winner. When using the active cluster wins rule, I/Os to the consistency
group continue at the cluster where the application was actively writing last (provided
there was only one such cluster. If dirty buffers exist in both clusters for the same
asynchronous consistency group, I/Os to the consistency group will suspend on both
clusters). In this way, the active cluster is the preferred cluster.
VPLEX Witness
VPLEX Witness, introduced with GeoSynchrony 5.0, is an optional external server that
is installed as a virtual machine. VPLEX Witness connects to both VPLEX clusters over
the management IP network. By reconciling its own observations with the information
reported periodically by the clusters, the VPLEX Witness enables the cluster(s) to
distinguish between inter-cluster network partition failures and cluster failures and
automatically resume I/O in these situations at the appropriate site. With
GeoSynchrony 5.0 VPLEX Witness affects only synchronous consistency groups in a
VPLEX Metro configuration, and only when the detach rules depict either cluster-1 or
cluster-2 as preferred for the consistency group (that is, VPLEX Witness does not
affect consistency groups where No Automatic Winner rule is in effect).
Without VPLEX Witness, if two VPLEX clusters lose contact, the consistency group
detach rules in effect define which cluster continues operation and which suspends
I/O as explained earlier. The use of detach rules alone to control which site is a
winner may add an unnecessary complexity in case of a site failure since it may be
necessary to manually intervene to resume I/O to the surviving site. VPLEX Witness
handles such an event dynamically and automatically. It provides the following
features:
Automatic load balancing between data centers
Active/active use of both data centers
Fully automatic failure handling
2
Based on management GUI options. CLI uses slightly different terms to specify the same rules.

In order for VPLEX Witness to be able to distinguish correctly between failure
conditions, it must be installed in an independent failure domain from either cluster
using distinct network interfaces to each. This will eliminate the possibility of a single
fault affecting both the cluster and the VPLEX Witness. For example, if the two clusters
of a VPLEX Metro configuration are deployed on two different floors of the same data
center, deploy the VPLEX Witness on a separate floor. On the other hand, if the two
clusters of a VPLEX Metro configuration are deployed in two different data centers,
deploy the VPLEX Witness in the third data center.
Cache vault
To avoid metadata loss or data loss (when write back caching is used) under
emergency conditions, VPLEX uses a mechanism called cache vaulting to safeguard
cache information to a persistent local storage.
Oracle Real Application Clusters

Oracle Real Application Clusters ) are an optional feature of Oracle Database
Enterprise Edition. Oracle RAC supports the transparent deployment of a single
database across a cluster of servers, providing fault tolerance from hardware failures
or planned outages. If a node in the cluster fails, the Oracle database continues
running on the remaining nodes. If more processing power is needed, new nodes can
be added to the cluster by providing horizontal scaling.
Oracle RAC supports mainstream business applications of all kinds. This includes
Online Transaction Processing (OLTP) and Decision Support System (DSS).
Oracle Automatic Storage Management

Oracle Automatic Storage Management (ASM) is an integrated database filesystem
and disk manager. ASM filesystem and volume management capabilities are
integrated with the Oracle database kernel. This reduces the complexity of managing
the storage for the database. In addition to providing performance and reliability
benefits, ASM can also increase database availability because disks can be added or
removed without shutting down the database. ASM automatically rebalances the files
across the disk group after disks have been added or removed.
In Oracle Database 11gR2, Oracle ASM and Oracle Clusterware have been integrated
into a single set of business and named Oracle Grid Infrastructure. This now provides
all the cluster and storage services required to run an Oracle RAC database. Oracle
ASM has also been extended to include support for Oracle Cluster Registry (OCR) and
voting files to be placed within ASM disk groups.
Oracle RAC on Extended Distance Clusters

Oracle RAC on Extended Distance Clusters is an architecture where servers in the
cluster reside in locations that are physically separate. Oracle RAC on Extended
Distance Clusters provides greater availability than local Oracle RAC. Oracle RAC on
Extended Distance Clusters provides extremely fast recovery from a site failure and
allows for all servers, in all sites, to actively process transactions as part of a single

database cluster. While this architecture creates great interest and has been
successfully implemented, it is critical to understand where this architecture best fits
especially in regards to distance, latency, and degree of protection it provides. The
high impact of latency, and therefore distance, creates some practical limitations as
to where this architecture can be deployed. This architecture fits best where the two
datacenters are located relatively close and where the costs of setting up direct
dedicated channels between the sites have already been taken.
Symmetrix VMAX
Symmetrix VMAX is built on the strategy of simple, intelligent, modular storage, and
incorporates a new Virtual Matrix™ interconnect that connects and shares resources
across all nodes, allowing the storage array to seamlessly grow from an entry-level
configuration into the world’s largest storage system. It provides the highest levels of
performance and availability featuring new hardware capabilities as seen in Figure 6.
2 – 16 director boards
Up to 2.1 PB usable
capacity
Up to 128 FC FE ports
Up to 64 FICON FE ports
Up to 64 GigE / iSCSI FE
ports
Up to 1 TB global memory
(512 GB usable)
48 – 2,400 disk drives
Enterprise Flash Drives
Figure 6. The Symmetrix VMAX platform 200/400 GB
FC drives 146/300/450/600
Symmetrix VMAX provides the ultimate scale-out
GB 15k platform.
rpm It includes the ability to
incrementally scale front-end and back-end(orperformance by adding processing
400 GB 10k rpm)
modules (nodes) and storage bays. Each processing module provides additional
SATA II drives 2 TB 7.2k
front-end, memory, and back-end connectivity.
rpm
Symmetrix VMAX also increases the maximum hyper size to 240 GB (64 GB on
Symmetrix DMX™). This allows ease of storage planning and device allocation,
especially when using Virtual Provisioning™ where the thin storage pool is already
striped and large hypers can be easily used.
Symmetrix VMAX TimeFinder product family
The EMC TimeFinder® family of local replication technology allows for creating
multiple, nondisruptive, read/writeable storage-based replicas of database and
application data. It satisfies a broad range of customers’ data replication needs with
speed, scalability, efficient storage utilization, and minimal to no impact on the

applications – regardless of the database size. TimeFinder provides a solution for
backup, restart, and recovery of production databases and applications, even when
they span Symmetrix arrays. TimeFinder is well integrated with other EMC products
such as SRDF and allows the creation of replicas on a remote target without
interrupting the synchronous or asynchronous replication. If a restore from a remote
replica is needed, TimeFinder and SRDF will restore data incrementally and in parallel,
to achieve a maximum level of availability and protection. The TimeFinder product
family supports the creation of dependent write-consistent replicas by using EMC
consistency technology, and replicas that are valid for Oracle backup/recovery
operations, as described later in the use cases.
Virtual Provisioning
Symmetrix thin devices are logical devices that you can use in many of the same ways
that Symmetrix devices have traditionally been used. Unlike traditional Symmetrix
devices, thin devices do not need to have physical storage preallocated at the time
the device is created and presented to a host (although in some cases customers
interested only in the thin pool wide striping and ease of management choose to fully
preallocate the thin devices). You cannot use a thin device until it has been bound to
a thin pool. Multiple thin devices may be bound to any given thin pool. The thin pool
is comprised of devices called data devices that provide the actual physical storage
to support the thin device allocations. Table 2 describes the basic Virtual Provisioning
definitions.
Table 2. Definitions of Virtual Provisioning devices
Device Description
Thin device A host-accessible device that has no storage directly associated with it.
Data Internal devices that when placed in a thin pool provide storage capacity to be used
devices by thin devices.
Thin pool A collection of data devices that provide storage capacity for thin devices.
Implementing VPLEX Metro with Oracle Extended RAC

Oracle Extended RAC deployment considerations in a VPLEX Metro
environment
EMC VPLEX breaks the physical barriers of data centers and allows users to access
data at different geographical locations concurrently. Oracle Extended RAC with
VPLEX Metro allows for transparent workload sharing between multiple sites
accessing a single database, while providing the flexibility of migrating workloads
between sites in anticipation of planned events such as hardware maintenance.
Furthermore, in case of an unplanned event that causes disruption of services at one
of the data centers, the failed client connections can be automatically restarted at the
surviving site.

Oracle Extended RAC and VPLEX Metro deployment guidelines
The following points describe some of the main deployment guidelines.
Oracle Clusterware and VPLEX Witness deployment

Often deployments of Oracle Extended RAC emphasize deploying a third site for one
of the Oracle clusterware voting disks (possible based on NFS). However in the
deployment of Oracle Extended RAC with VPLEX Metro, it is the VPLEX Witness alone
that is deployed in an additional fault domain (third site in the case of multi-site
deployment) as described in the VPLEX Witness section earlier:
In the case of Oracle interconnect partitioning alone (not a true site failure, and no
affect on VPLEX interconnect) then Oracle clusterware will reconfigure based on
node majority and access to the voting disk. No need for a third site for Oracle
clusterware.
In the case of VPLEX interconnect partitioning (or a true site failure) VPLEX
immediately suspends IOs at both clusters while VPLEX Witness resolves which of
them should continue to service IOs (based on detach rules and site availability).
The Oracle cluster nodes will therefore have access to voting disks only where
VPLEX resumes IO, and Oracle clusterware will therefore reconfigure the cluster in
accordance. Again, no need for Oracle clusterware to use a third site.
Oracle Clusterware deployment

Oracle Cluster Ready Service (CRS) is deployed on VPLEX distributed volumes alone
(no third sites) as described in the previous section. In Oracle database, release
11gR2 CRS was merged with ASM to create the grid infrastructure. Therefore, the first
ASM disk group is created at the time of CRS installation. In prior database releases,
the Oracle clusterware uses either raw devices or a clustered filesystem:
When ASM is used for Oracle clusterware (starting with Oracle 11gR2) it is
recommended to create a unique disk group for CRS alone, for example: +GRID
(that is, no database content such as log or data files will be placed there). The
+GRID disk group will benefit from using Normal or High redundancy. This way
Oracle will create multiple voting disks (instead of only one such as the case if
CRS ASM disk group was using External redundancy3). Because no database
content is included, the size of the distributed VPLEX devices that are used for this
disk group can be relatively very small.
The recommended approach when EMC storage is used behind VPLEX, such as
VMAX or VNX, or when VPLEX RAID protection is used, is to set all other ASM disk
groups with external redundancy. This will provide adequate protection for the
ASM members, based on VPLEX or EMC storage array RAID.
3
Starting with Oracle 11gR2, the number of CRS voting disks is determined automatically by the ASM redundancy level, such
as External redundancy implies 1 voting disk, Normal redundancy implies 3 voting disks, and High redundancy implies 5
voting disks.

Since all Oracle cluster nodes require access to all CRS and database devices,
both CRS and Oracle database should use only the VPLEX distributed volumes,
whether ASM is used, raw devices, or a clustered filesystem.
To align with timeout values associated with VPLEX cluster and Witness resolution
of failure states, it is recommended to modify the CRS CSS Miscount from the
default of 30 seconds to 45 seconds. This can be done by using the following
commands:
# crsctl get css miscount
# crsctl set css miscount 45
No other timeout value changes are necessary (that is, such as crs disk timeout).
Additional notes
On x86-based server platform, ensure that partitions are aligned. VPLEX requires
alignment at 4 KB offset; however, if Symmetrix is used align at 64 KB (128 blocks)
offset (which is natively aligned at 4 KB boundary as well):
On Windows diskpar or diskpart can be used. On Linux fdisk or parted can be
used.
An example of aligning partition to 64 KB offset using fdisk is shown later in the
section: Create partitions on PowerPath devices.
Oracle Extended RAC and VPLEX Metro protection from unplanned downtime
The combination of Oracle Extended RAC and VPLEX Metro provides improved
availability and resiliency to many failure conditions and therefore increased the
availability of mission-critical database and applications.
Table 3 summarizes the list of failure scenarios and the best practices that will make
the database able to continue operations in each of them. Note that the list does not
cover failover to a standby system (such as Oracle Data Guard, RecoverPoint, SRDF,
and so on). EMC VPLEX with GeoSynchrony 5.0 Product Guide provides more
information regarding VPLEX connectivity best practices.

Table 3. Summary of VPLEX Metro and Oracle Extended RAC improved resiliency
Resiliency to host and site failure scenarios
Failure Oracle database single server Oracle RAC Oracle Extended RAC
(non-RAC) (not extended) with VPLEX Metro
Host HBA port Each host should have

failure more than a single path to the
Same as single server Same as single server
storage. Use multiple HBA ports
(initiators).
Use multipathing software
(such as EMC PowerPath®) for
automatic path failover and
load balancing.
Ideal SAN connectivity will use
redundant switches where the
HBA ports (initiators) will
spread across them.
Host hardware Downtime implied until host Oracle RAC provides Same as Oracle RAC
failure or crash and application can resume database resiliency for a
operations. single server failure by
performing automatic
instance recovery and
having other cluster
nodes ready for user
connections.
Configure Oracle
Transparent Application
Failover to allow
sessions to automatically
failover to a surviving
cluster node.
Lab/building/site Downtime implied until host Downtime implied until By installing VPLEX
failure and application can resume host and application can clusters and Witness in
operations. resume operations. independent failure
domains (such as
another building or
site) it becomes
resilient to a lab,
building, or site
failures.
The VPLEX cluster in
the failure-domain not
affected by the
disaster will continue
to serve I/Os to the
application.

Use Oracle Transparent
Application Failover to
allow automatic user
connection failover to
the surviving cluster
nodes.
Resiliency to database/network related failure scenarios
Database instance Downtime implied until Oracle RAC provides

crash or public instance can be brought up database resiliency for a
Same as Oracle RAC
network disconnect again or public network single server failure by
reconnected. performing automatic
instance recovery and
having other cluster
nodes ready for user
connections.
Configure Oracle
Transparent Application
Failover to allow sessions
to automatically failover
to a surviving cluster
node.
RAC Cluster N/A RAC natively and

interconnect automatically handles by
Same as Oracle RAC
partitioning cluster reconfiguration.
Cluster nodes will
compete for voting disk
access.
Resiliency to Storage failure scenarios
(non-RAC) (extended) with VPLEX Metro
Front-end port SAN connectivity should

failure include multiple storage front-
end ports, ideally across
Symmetrix directors. If using
Symmetrix with multiple
engines connect to ports on
different engines as well to gain
even higher protection..
Physical drive Use storage RAID

failure protection. Symmetrix storage
uses RAID protection where
RAID1 and RAID5 protect from a
single disk failure within a RAID

group and RAID6 protects from
two disks failures within a RAID
group. In either case the
application will continue to run
undisturbed.
If a drive starts failing
Symmetrix hot-spare drive will
copy its data and EMC
Enginuity™ will initiate Call-
Home to inform EMC support
immediately.
Storage array Symmetrix components are

components, fully redundant, including
including director mirrored cache that is also
board (cache, IO) persistent (uses vaulting in the
case of elongated power
failure), redundant directors
and power supplies.
Symmetrix data is T10 DIF
protected from the time it
enters the storage until it
leaves it.
Loss of connectivity Downtime implied until storage Downtime implied until VPLEX Metro
to storage array array connectivity can be storage array synchronous
resumed. connectivity can be consistency group
resumed. continues to serve
I/O’s at both sites,
even if one of the
storage arrays is not
available.
Oracle clusterware
would not know about
the storage
unavailability as VPLEX
cluster continues to
service all I/Os.
Resiliency to VPLEX failure scenarios
Failure Oracle Database Single Server Oracle RAC Oracle Extended RAC
Front-end port SAN connectivity should

include multiple storage front-
end ports, ideally across
Symmetrix directors. If using
VPLEX with multiple engines
connect to ports on different
engines as well to gain even
higher protection.

Use multipathing software
(such as PowerPath) for
automatic path failover and
load balancing.
Ideal SAN connectivity will use
redundant switches with
connectivity to multiple VPLEX
front-end ports.
Back-end port In a similar fashion to VPLEX

front-end ports, also use
redundant switch connectivity
with VPLEX back-end ports as
they connect to the storage
array.
VPLEX hardware VPLEX components are fully

component redundant, including persistent
cache (uses vaulting in the case
of elongated power failure),
redundant directors and power
supplies.
VPLEX Interconnect N/A N/A If both sites are still

partition available the VPLEX
Preferred Cluster detach
rules will determine
which cluster resumes
I/Os and which
suspends, without
downtime for host
connected to the
surviving Cluster.
VPLEX cluster Downtime implied until VPLEX Downtime implied until VPLEX Witness will
unavailable cluster connectivity can be VPLEX cluster allow I/O to resume at
resumed. connectivity can be the surviving VPLEX
resumed. cluster. RAC cluster
nodes connected to
that VPLEX cluster will
resume operations
without downtime.
Use Oracle Transparent
Application Failover
(TAF) to allow
automatic user
connection failover to
the surviving cluster
nodes.
Although the architecture of VPLEX is designed to support concurrent access at

multiple locations, the current version of the product supports a two-site
configuration separated by synchronous distance with a maximum round trip latency

of 5 ms between the two sites. In addition, the VPLEX Metro solution requires
extension of a VLAN to different physical data centers. Technologies such as Cisco’s
Overlay Transport Virtualization (OTV) can be leveraged to provide the service. EMC
VPLEX Architecture and Deployment: Enabling the Journey to the Private Cloud
TechBook, available on EMC Powerlink®, provides further information on EMC VPLEX
Metro configuration.
VPLEX Metro with Oracle Extended RAC lab configurations and

tests
Lab configuration and setup
The following section describes the technologies and components used in the test
cases documented in this paper.
Physical environment
Figure 7 illustrates the overall physical architecture of the Oracle Extended RAC
configuration deployment that was used for the tests shown in this paper. The Oracle
Extended RAC consisted of four Oracle RAC cluster nodes, two at each simulated data
center (Site A and Site B).

Figure 7. VPLEX Metro configuration for Oracle
The hardware setup included setting up Fibre Channel connectivity among the hosts,
VPLEX clusters, and Symmetrix VMAX storage arrays to the redundant switches. In
each site, the hosts are zoned to their local VPLEX cluster front-end ports, and the
storage to the VPLEX cluster back-end ports. The VPLEX clusters are zoned to each
other (through the distance simulator in this case). The Oracle RAC interconnect
between sites used the same distance simulator to experience the same latency
overhead as the VPLEX interconnect. The distance simulator used 2 x 1 GigE links
with WAN compression and Ciena switches for protocol change.
Table 4 shows the VPLEX, hosts, and Symmetrix storage array hardware details. They
included 2 x 2-engine VPLEX clusters setup in Metro configuration, and 2 x 1-engine
Symmetrix VMAX arrays providing local storage to each VPLEX cluster. The 4 x RAC
nodes consisted of Dell 2950 dual core servers with 16 GB of cache. It should be
noted that the focus of the tests was on generating high IO workload rather than
making use of lots of RAM to achieve high transaction rate numbers. VPLEX Witness
was set and connected to each VPLEX cluster.
Table 5 shows the distance simulation hardware. It included Dell PowerConnect
6224F Fiber Ethernet Switch that converts Gigabit IP network into Fibre Channel
network. Ciena CN 2000 network manager provides virtual optical networking with
inputs of Fibre Channel, FICON, or GbE. It supports both data compression and
Dynamic Bandwidth Allocation (DBA) to provide port level quality of service to client
connections. Empirix PacketSphere Network Emulator provided a controlled network

emulation for IP network impairments, such as network latency, for full-duplex,
Gigabit wire speed network traffic.
Table 4. Extended RAC hardware environment
Hardware Quantity Release and configuration

EMC VPLEX Metro 2 VPLEX Metro with GeoSynchrony 5.0.1
Two engine and four directors on each
cluster
VPLEX Witness 1 Dell R900 running VPLEX Witness
virtual machine (VM)
Symmetrix VMAX 2 Single engine VMAX with Enginuity
5875, 112 x 450 GB/15k FC drives
using Virtual Provisioning
Dell 2950 server (RAC 4 2 x dual core, 16 GB RAM
nodes)
Emulex Express (2 HBA ports 4
used per server)
The main guidelines during the connectivity step are to maximize hardware
redundancy such as using two switches, more than a single HBA, multipath for
dynamic path failover, and load balancing.
Table 5. Distance simulation hardware environment
Hardware Quantity Release and configuration

Dell PowerConnect 6224F 1 Support up to 4 x 10 Gigabit fiber and
Fiber Ethernet Switch 2 x 10GBase-T copper Ethernet uplinks
Ciena CN 2000 Network 1 Ciena ON-Center CN 2000 Network
Manager Manager 5.0.1
Empirix PacketSphere 1 Empirix Network Emulator NE XG
Network Emulator 1.0B8, 2 1-GbE network emulators
Dell 2950 server 4 2 x dual core, 16 GB RAM
Table 6 describes the host software used.

Table 6. Host software
Software Release
Server OS Oracle Enterprise Linux Release 5 Update 4
x86_64
EMC PowerPath Version 5.5 for Linux x86_64
Oracle Oracle Clusterware 11g R2 (11.2.02) and Oracle
Database 11g R2 (11.2.0.2) for Linux x86-64

Storage setup and device allocation planning
Symmetrix VMAX Virtual Provisioning and storage device configuration

Table 7 describes Symmetrix VMAX device configuration with storage Virtual
Provisioning and volume layout for VPLEX Metro with Oracle Extended RAC testing
environment.
This configuration placed the Oracle data and log files in separate thin pools which
allowed each to use distinct RAID protection. In this configuration, data files were
placed in the RAID5 protected thin pool and redo logs in the RAID1 protected thin
pool:
Symmetrix RAID5 protected thin pool offers a good combination of protection,
performance, and capacity utilization benefits for the data files with optimized
writes and rotating parity, therefore being a good choice for data files. RAID1
protected thin pool in certain cases may offer a slight improvement in availability
and performance over RAID5 and therefore was used for the log files. Note that the
same group of physical disks was shared by both thin pools to allow full sharing
of physical resources.
It should be noted that a different configuration, focused on simplicity rather than
pure storage performance/availability optimization, could be deployed where
both data and log files share the same thin pool and RAID protection (be it RAID1,
RAID5, or RAID6).
Multiple ASM disk groups were used in the configuration:
+GRID: As discussed earlier in Oracle Clusterware deployment section, when ASM
is used for Oracle clusterware (starting with Oracle 11gR2) it is recommended to
create a unique disk group for CRS alone, for example: +GRID. This approach of
separating the clusterware LUNs from the database is useful when there are plans
to use storage technologies such as clones or snapshots to create additional
database copies for repurposing, backup, and so on. These copies will not include
the clusterware LUNs. It is also helpful for DR solutions such as RecoverPoint and
SRDF since the replicated ASM disk groups exclude the clusterware LUNs, and will
be mounted to an already configured CRS stack at the DR target site.
Separating +DATA, +LOG, and +FRA ASM disk groups allow the storage technology
to be used for offloading backups from production. During hot backup process the
+DATA and +FRA diskgroups will be cloned at different times. In addition, remote
replications such as RecoverPoint and SRDF natively create a database restartable
replica. Restartable replicas do not access archive logs during crash or instance
recovery and therefore the archive logs (+FRA disk group) do not need to be part of
the replication.
Temp files often can be included in the +DATA disk group. The only reason they
used a separate ASM disk group in the tests was just for monitoring purposes and
it is not a specific deployment recommendation.

Table 7. Storage and database devices configuration and assignment
Thin devices (LUNs)
ASM disk group Thin pool Thin Data devices

and LUNs binding devices
assignment
Oracle RAC +Grid ASM disk 5 x 20 GB thin

Grid/ASM group LUNs (15F:163) 56 x 30GB
Redo_Pool
Instance RAID1
+REDO ASM disk 5 x 20 GB thin
Redo_Pool
group LUNs (164:168)
25 x 60 GB thin 56 x 230 GB
Database: +DATA: ASM disk LUNs (1A5:1B4) RAID5 (3+1)
Data_Pool
Name: group
(C5:FC)
ERPFINDB
Size: 1 TB
+TEMP: ASM disk 6 x 50 GB thin
Num. LUNs: 38
group LUNs (17D:182)
+FRA: ASM disk 2 x 50 GB

56 x 60 GB
group thin LUNs
RAID5 (3+1)
Temp_Pool
VPLEX Meta 2 x 2 x 80 GB (8D:C4)
devices thin LUNs
(2E5:2E8)
VPLEX
2 x 50 GB thin
VPLEX Log
LUNs (2E9:2EA)
Symmetrix VMAX storage provisioning for VPLEX Metro system

The following steps were followed to provision storage from the Symmetrix VMAX
system to a VPLEX virtual storage environment, which is basically the same as to
provision storage to physical or virtual servers. The process described assumes this is
the first time storage is being provisioned from the Symmetrix VMAX array to the
VPLEX Metro system. It also assumes that the VPLEX have been zoned to front-end
ports of the Symmetrix VMAX storage array. The Symmetrix actions should be
executed from a management host connected through gatekeeper to the Symmetrix,
or using a Symmetrix Management Console client. The following list shows the setup

activities with focus on CLI although they can be easily executed using Symmetrix
Management Console as well.
Step Action
1 Create Symmetrix devices on both local and remote arrays using either
Symmetrix Management Console or the Solutions Enabler command line
interface.
2 a. Create a Symmetrix storage group by executing the following command:
b. symaccess –sid <symm_id> –name <group name> –type storage devs
create
c. For example, the command:

d. symaccess –sid 191 –name VPLEX1_Storage_Group –type storage devs
15F:168 create
e. creates the storage group, Storage_Group_Test.

3 f. Create a Symmetrix port group. The command:
symaccess –sid 191 -name VPLEX1_Port_Group -type port –dirport
<Dir>:<Port> create
g. For example, the command:

symaccess –sid 191 -name VPLEX1_Port_Group -type port –dirport 7E:1
create
creates the port group, Port_Group_Test.

4 h. Create a Symmetrix initiator group where the VPLEX back-end ports’ WWNs are
the “host” initiators to the Symmetrix initiator group. The creation of the initiator
group, Initiator_Group_Test, can be achieved by running the following command:
symaccess –sid 191 –name VPLEX1_Initiator_Group –type init –wwn
<WWN> create
i. For example, the command:

symaccess –sid 191 –name VPLEX1_Initiator_Group –type init –wwn
500014426011ee10 create
5 j. Create the Symmetrix masking view to group the storage, port, and initiator
groups:
k. symaccess –sid 191 create view –name VPLEX1_View –storgrp
VPLEX1_Storage_Group –portgrp VPLEX1_Port_Group –initgrp
VPLEX1_Initiator_Group
6 l. Repeat steps 1–5 for storage provisioning the second VPLEX system (VPLEX2) to
the second VMAX array (sid 219)

VPLEX Metro setup
VPLEX Metro cluster setup steps
Figure 8 lists the main tasks that are required for VPLEX Metro setup.
Figure 8. Overview of VPLEX Metro setup tasks
Note: You must set up both VPLEX Metro clusters as described. You cannot set each
cluster up individually and then join them later.

Set up Metro-Plex cluster connectivity
Metro-Plex connectivity is the communication between clusters in a Metro-Plex. The
two key components of Metro-Plex communication are FC (FCIP, DWDM) and VPN
WAN. FC MAN is the Fibre Channel connectivity between directors of each cluster and
the VPN is connectivity between management servers for management purposes. A
Metro-Plex should be set up with redundant (dual fabrics) and completely
independent Fibre Channel connectivity between clusters for inter-cluster
communication. This provides maximum performance, fault isolation, fault tolerance,
and availability. Figure 9 is an example of zoning inter-cluster WAN connectivity.
Figure 9. Zoning example of VPLEX inter-cluster WAN connectivity
Checking cluster connectivity

To check for FC MAN connectivity, log in to the VPLEX CLI and run the following command:
An example:
VPlexcli:/> ll **/hardware/ports/
/engines/engine-1-1/directors/director-1-1-A/hardware/ports:
Name Address Role Port Status
------- ------------------ --------- -----------
A2-FC00 0x500014426011ee20 wan-com up
A2-FC01 0x500014426011ee21 wan-com up
A2-FC02 0x500014426011ee22 wan-com up
A2-FC03 0x500014426011ee23 wan-com up
/engines/engine-1-1/directors/director-1-1-B/hardware/ports:
------- ------------------ --------- -----------
B2-FC00 0x500014427011ee20 wan-com up

B2-FC01 0x500014427011ee21 wan-com up

B2-FC02 0x500014427011ee22 wan-com up
B2-FC03 0x500014427011ee23 wan-com up
/engines/engine-2-1/directors/director-2-1-A/hardware/ports:
------- ------------------ --------- -----------
A2-FC00 0x5000144260168220 wan-com up
A2-FC01 0x5000144260168221 wan-com up
A2-FC02 0x5000144260168222 wan-com up
A2-FC03 0x5000144260168223 wan-com up
/engines/engine-2-1/directors/director-2-1-B/hardware/ports:
------- ------------------ --------- -----------
B2-FC00 0x5000144270168220 wan-com up
B2-FC01 0x5000144270168221 wan-com up
B2-FC02 0x5000144270168222 wan-com up
B2-FC03 0x5000144270168223 wan-com up
To check FC MAN link status, run cluster summary command.
An example:
VPlexcli:/> cluster summary

Clusters:
Name Cluster ID Connected Expelled Operational Status Health State
--------- ---------- --------- -------- ------------------ ------------
cluster-1 1 true false ok ok
cluster-2 2 true false ok ok
Islands:
Island ID Clusters
--------- --------------------
1 cluster-1, cluster-2
VPLEX Metro host connectivity

To ensure the highest level of connectivity and availability to Oracle Extended RAC
even during abnormal operations for connecting Oracle RAC Servers to EMC VPLEX,
each Oracle RAC Server in the Oracle Extended RAC Infrastructure environment should
have at least two physical HBAs, and each HBA should be connected to front-end
ports on different directors on EMC VPLEX. This configuration ensures continued
availability of the Oracle RAC Server even if one of the front-end ports of the EMC
VPLEX goes offline for either planned maintenance events or unplanned disruptions.

When a single VPLEX engine configuration is connected to an Oracle Extended RAC
environment, each HBA should be connected to the front-end ports provided on both
the A and B directors within the VPLEX engine. Connectivity to the VPLEX front-end
ports should consist of first connecting unique hosts to port 0 of each I/O module
emulating the front-end directors before connecting additional hosts to the remaining
ports on the I/O module. If multiple VPLEX engines are available, the HBAs from the
Oracle RAC Servers should be connected to different engines.
The connectivity from the VPLEX engines to the storage arrays should follow the best
practices recommendation for the array. A detailed discussion of the best practices
for connecting the back-end storage is beyond the scope of this paper. The EMC
VPLEX Architecture and Deployment: Enabling the Journey to the Private Cloud
TechBook provides more information.
VPLEX Metro administration

Administration of VPLEX Metro can be done primarily through the Management
Console, although the same functionality exists with VPLEX CLI. On authenticating to
the secure web-based GUI, the user is presented with a set of on-screen configuration
options, listed in the order of completion. The EMC VPLEX Management Console
online help provides more information about each step in the workflow. The following
table summarizes the steps to be taken from the discovery of the arrays up to the
storage being visible to the host.
Step Action
1 Discover available storage
VPLEX Metro automatically discovers storage arrays that are connected to the back-end
ports. All arrays connected to each director in the cluster are listed in the Storage Arrays
view.
2 Claim storage volumes
Storage volumes must be claimed before they can be used in the cluster (with the
exception of the metadata volume, which is created from an unclaimed storage volume).
Only after storage volume is claimed, it can be used to create extents, devices, and then
virtual volumes.
3 Create extents
Create extents for the selected storage volumes and specify the capacity.
4 Create devices from extents
A simple device is created from one extent and uses storage in one cluster only.
5 Create a virtual volume

Create a virtual volume using the device created in the previous step.
6 Register initiators
When initiators (hosts accessing the storage) are connected directly or through a Fibre
Channel fabric, VPLEX Metro automatically discovers them and populates the Initiators
view. Once discovered, you must register the initiators with VPLEX Metro before they can
be added to a storage view and access storage. Registering an initiator gives a

meaningful name to the port’s WWN, which is typically the server’s DNS name, to allow
you to easily identify the host.
7 Create a storage view
For storage to be visible to a host, first create a storage view and then add VPLEX Metro
front-end ports and virtual volumes to the view. Virtual volumes are not visible to the hosts
until they are in a storage view with associated ports and initiators.
8 Create consistency group
Create a consistency group on both VPLEX systems, and add all virtual
volumes assigned for Oracle Extended RAC ASM devices, including Grid and
ASM devices for Oracle RAC database, that require write-order consistency to
the consistency group.
Figure 10 shows the GUI interface for logical layout and provisioning storage from
EMC VPLEX.
Figure 10. EMC VPLEX GUI management interface

The browser-based management interface, as seen in Figure 10, schematically shows
the various components involved in the process. Storage from EMC VPLEX is exposed
using a logical construct called “Storage View” that is a union of the objects,
“Registered initiators,”, “VPLEX ports,” and “Virtual Volume”. The “Registered
initiators” object lists the WWPN of the initiators that need access to the storage. In
the case of an Oracle VM Server environment, the “Registered initiators” entity

contains the WWPN of the HBAs in the Oracle VM Servers connected to the EMC
VPLEX. The object “VPLEX ports” contains the front-end ports of the VPLEX array
through which the “Registered initiators” access the virtual volumes. The “Virtual
Volume” object is a collection of volumes that are constructed from the storage
volumes that are provided to the EMC VPLEX from the back-end storage arrays. It can
be seen from the inset in the bottom left corner of Figure 10 that a virtual volume is
constructed from a “Device” that in turn can be a combination of different devices
built on top of an abstract entity called an “Extent”. The figure also shows that an
“Extent” is created from the “Storage Volume” exposed to the EMC VPLEX. However,
to leverage array-based replication technologies, we map each storage device from
VMAX in its entire (one-to-one mapping) pass-through configuration (device capacity
= extent capacity = storage volume capacity) to VPLEX and has a RAID 0 (single extent
only) VPLEX device geometry. By doing so, the underlying storage devices are left
untouched by VPLEX and the back-end array LUN replication technology, such as
TimeFinder/Clone and TimeFinder/Snap, continues to function normally.
Also shown in Figure 10 in the bottom right corner inset are the seven steps that are
required to provision storage from EMC VPLEX. The wizard supports a centralized
mechanism for provisioning storage to different cluster members in case of EMC
VPLEX Metro. The first step in the process of provisioning storage from EMC VPLEX is
the discovery of the storage arrays connected to it. This step needs to be rarely
executed since the EMC VPLEX proactively monitors for changes to the storage
environment. The second step in the process is the “claiming” of storage that has
been exposed to EMC VPLEX. The process of claiming the storage creates the object’s
“Storage Volume” that is shown in Figure 10. The Create Storage View wizard enables
you to create a storage view and add initiators, ports, and virtual volumes to the view.
Once all the components are added to the view, it automatically becomes active.
When a storage view is active, hosts can see the storage and begin I/O to the virtual
volumes. After creating a storage view, you can only add or remove virtual volumes
through the GUI. To add or remove ports and initiators, use the CLI. The EMC VPLEX
CLI Guide provides comprehensive information above VPLEX Metro commands.
VPLEX Metro with VPLEX Witness

VPLEX Witness is installed as a closed virtual machine deployed in a failure domain
separate from either of the VPLEX clusters (to eliminate the possibility of a single fault
affecting both the cluster and the VPLEX Witness). VPLEX Witness connects to both
VPLEX clusters over the management IP network. By reconciling its own observations
with the information reported periodically by the clusters, the VPLEX Witness enables
the cluster(s) to distinguish between inter-cluster network partition failures and
cluster failures and automatically resume I/O in these situations.
Host and Oracle setup
Multipathing software setup

As part of the best practice for the connectivity of Oracle RAC Servers to VPLEX, each
Oracle RAC Server should have two HBA ports with each port connected to a separate

FC switch for increased availability. In such a configuration, the host must use a
multipathing solution to handle multiple paths to the same storage device for the
purpose of providing high availability, load balancing, and live migration. You can
either install EMC PowerPath for Oracle RAC Server as the multipathing solution, or
use the Linux native multipath solution, device mapper, to configure Oracle RAC
Servers. For Oracle Extended RAC in an EMC VPLEX Metro configuration, as described
in this paper, we installed EMC PowerPath 5.5 on four physical servers.
Install PowerPath rpm on each host

[root@ RAC NODE 1: licoc039 ]rpm -i EMCpower.LINUX-5.5.0.00.00-
275.RHEL5.x86_64.rpm
A reboot may be required the first time after PowerPath is installed for the host to
register the /dev/emcpower pseudo devices.
Install a PowerPath license on each host
[root@ RAC NODE 1: licoc039 ]emcpreg –add <key>
Configure PowerPath on each host
[root@ RAC NODE 1: licoc039 ] powermt config
[root@ RAC NODE 1: licoe039 ] powermt display
…
Pseudo name=emcpowerk
Invista ID=FNM00100600231
Logical device ID=6000144000000010A002636D3C679C6A
state=alive; policy=ADaptive; priority=0; queued-IOs=0
==============================================================================
---------------- Host --------------- - Stor - -- I/O Path - -- Stats ---
### HW Path I/O Paths Interf. Mode State Q-IOs Errors
==============================================================================
1 lpfc sdaq 08 active alive 0 0
2 lpfc sdbu 00 active alive 0 0
2 lpfc sdcy 08 active alive 0 0
1 lpfc sdm 00 active alive 0 0
Match PowerPath pseudo device names across Oracle RAC Server nodes
To match the PowerPath pseudo device names across Oracle RAC Server nodes, EMC
recommends to use the PowerPath utility emcpadm. The utility allows to export the
mapping from one host and to import it to another. It also allows the renaming of
pseudo devices one at a time if necessary.
<source host> emcpadm export_mapping -f <mapping_file_name>

Then copy that file to the other hosts. Shut down any application using storage
devices, unmount any filesystem, or export any LVM volumes and run:
<target host> emcpadm check_mapping [-v] -f <mapping_file_name>
<target host> emcpadm import_mapping -f <mapping_file_name>
Create partitions on PowerPath devices

EMC strongly recommends to align x86-based server platform partitions offset to a 64
KB when Symmetrix is used. If another storage array is used it may have different
requirements although the size should always be aligned on 4 KB offset to match
VPLEX block size. Figure 11 shows alignment at 64 KB to create the partitions on a
PowerPath device, simply start fdisk as shown below, and once the partitions are
created, type “x” to enter Expert mode. Type “p” to show (print) the current partition
table, including the offset in block units. Type “b” to change any partition offset. For
example, move partition 1 from its default offset of 32 blocks to 128. Since each
block is 512 bytes, then 128 x 512 bytes = 64 KB offset. If more than one partition is
created on the LUN, verify that the rest of the partitions are aligned, or follow a similar
step to change their offset to a number that is 128 blocks (64 KB) aligned.
M
B
Reserved Default start New start
R (63 blocks) for partition for partition
1 1
64 KB 64 KB
Figure 11. Partition alignment to Symmetrix track size boundary (64 KB)
In this example, one partition is created for a PowerPath device that is going to be
used as an Oracle ASM device.
[root@licoc091 ~]# fdisk /dev/emcpowerd
Command (m for help): n
Command action
e extended
p primary partition (1-4)
p
Partition number (1-4): 1
First cylinder (1-52218, default 1): [ENTER]
Using default value 1
Last cylinder or +size or +sizeM or +sizeK (1-52218, default 52218):
[ENTER]
Command (m for help): p
Disk /dev/emcpowerd: 54.7 GB, 54755328000 bytes

64 heads, 32 sectors/track, 52218 cylinders
Units = cylinders of 2048 * 512 = 1048576 bytes

Device Boot Start End Blocks Id System
/dev/emcpowerd1 1 52218 53471168 83 Linux
Align partitions on a 64 KB boundary (128 blocks).

Command (m for help): x (going into export mode)
p (print partition table)

Expert command (m for help):
Note that partition 1 starts at 32 blocks
Disk /dev/emcpowerk: 64 heads, 32 sectors, 52218 cylinders
Nr AF Hd Sec Cyl Hd Sec Cyl Start Size ID

1 00 1 1 0 63 32 1023 32 106942336 83
2 00 0 0 0 0 0 0 0 0 00
3 00 0 0 0 0 0 0 0 0 00
4 00 0 0 0 0 0 0 0 0 00
Expert command (m for help): b (move partition start)
Partition number (1-4): 1
New beginning of data (32-2002943, default 32): 128
Expert command (m for help): p

Disk /dev/emcpowerk: 64 heads, 32 sectors, 52218 cylinders
Nr AF Hd Sec Cyl Hd Sec Cyl Start Size ID

1 00 1 1 0 63 32 1023 128 106942336 83
2 00 0 0 0 0 0 0 0 0 00
3 00 0 0 0 0 0 0 0 0 00
4 00 0 0 0 0 0 0 0 0 00
Expert command (m for help): w

The partition table has been altered!
Calling ioctl() to re-read partition table.

Syncing disks.
[root@licoc091 ~]#
After the partitions were created ensure the other nodes recognize them. It may be necessary to
run the fdisk command on each other node and write (“w”) the partition table. Alternatively, a
rescan of the SCSI bus or reboot of the other nodes will refresh the information as well.
Install Oracle and set up RAC database

The following table summarizes the steps to be taken to configure Oracle server
nodes for Oracle Grid Infrastructure and ASM database installation, according to
Oracle 11gR2.0.2 Grid Infrastructure Installation Guide for Linux and 11g Release
2.0.2 Database Installation Guide for Linux. The installation guide provides further
details.
Step Action
1 Configure server nodes private network, OS /etc/hosts file, OS kernel parameters and
edit the /etc/sysctl.conf file.

2 Create Oracle user groups and accounts for installing and maintaining Oracle 11gR2 on
each RAC nodes.
3 Update Boot Script (/etc/rc.d/rc.local)to set Oracle permissions for devices
designated for Oracle CRS and ASM.
4 Set up ssh for Oracle user on each RAC nodes.
5 Set the shell limits for Oracle user in the /etc/security/limits.conf file.
6 Modify the /etc/pam.d/login file and the /etc/profile file accordingly.
7 Install additional required OS packages for Oracle installation.
8 Install Oracle 11g R2.0.2 clusterware for grid infrastructure and create ASM instance for
Oracle CRS cluster on all the nodes.
9 Install Oracle 11g R2.0.2 database software.
10 Configure Oracle Cluster Synchronization Services (CSS) for ASM.
11 Create additional ASM disk groups for Oracle database.
12 Create an Oracle database with size and init parameters to meet the desired database
workload and performance requirements.
13 Register Database with CRS.
OLTP database workload tests

We used a standard Oracle OLTP workload (70/30 random-read/write ratio,
respectively) to illustrate that VPLEX Metro cluster provides Oracle Extended RAC with
high performance and workload resiliency over Metro distance of 100 km. As
mentioned in the implementation section of the paper, our Oracle Extended RAC with
VPLEX Metro test environment consists of two local RAC nodes and two remote RAC
nodes, the interconnect between VPLEX clusters and RAC nodes is affected by Metro
distance with Empirix PacketSphere Network Emulator for network latency. As shown
in Figure 12 and Figure 13, a 16-drivers workload was carried out from each one of
Oracle RAC nodes, OLTP workload transactions rates (transactions per minute) were
recorded for single node, two nodes, three nodes, or four nodes workload. The
workloads were increased with the addition of each node to show the scalability
benefits of RAC. The transaction rate increased proportionally as workload increase
with addition of more RAC nodes (16, 32, 48, and 64 workload drivers with 1–4
nodes, respectively). Similar transaction rate increases were shown for VPLEX Metro
with Oracle RAC at 0 km distance, as well as at 100 km Metro distance. Furthermore,
VPLEX Metro is capable of delivering high performance with approximately 85–
90percent of transaction rate for Oracle Extended RAC at Metro distance of 100 km, in
comparison with baseline, achieved at 0 km distance. In addition, VPLEX Metro is
capable of achieving over 90percent of workload performance for Oracle Extended
RAC at Metro distance of 50 km or less (data not shown). Therefore, VPLEX Metro
provides Oracle Extended RAC with high I/O performance and resiliency for OLTP
workload where the two datacenters are located at a Metro distance of 100 km. Tests
were performed also with 500 km distance (5 ms RTT) showing similar scalability
although with relative lower transaction rate due to increase latency. Overall the
solution proved both VPLEX and RAC ability to increase application performance and
provide higher availability at the same time. Note that since the OLTP benchmark was
indeed completely random, no block contention was encountered. In deployments
with real customer workload, the DBA should pay attention (as they always do) to

potential block contention across cluster nodes that, especially if occurs between
remote nodes, may reduce the overall transaction rate.
25000
Distance: 0 km (baseline)
Transaction Rate
20000
15000
10000
5000
0
1 2 3 4
Number of RAC nodes with active workload
Figure 12. Transaction rate at 0 km distance
25000
Distance: 100 km (1ms RTT)
Transaction Rate
20000
15000
10000
5000
0
1 2 3 4
Number of RAC nodes with active workload
Figure 13. Transaction rate at 100 km distance
Failure conditions tests

The environment was tested successfully for many failure scenarios driven by the
project team, E-LabTM (EMC quality and qualification organization) and based on
Oracle RAC engineering provided test plan. All the tests completed successfully with
the expected result. All the tests were performed while transaction workload was
running (when applicable). A very partial list of the tests performed includes:
RAC interconnect partitioning (‘split-brain’) while VPLEX Metro infrastructure is
intact.

RAC interconnect and VPLEX Metro interconnect partitioning.
A single storage system disconnect without application downtime.
Site failure simulation where the surviving site continues running workload.
Host connectivity loss for different RAC nodes without affecting the rest of the
cluster.
ASM rebalance, storage, and VPLEX config changes and software updates.
Conclusion
EMC VPLEX running the EMC Metro or GeoSynchrony operating system is an
enterprise-class SAN-based federation technology that aggregates and manages
pools of Fibre Channel attached storage arrays that can be either collocated in a
single data center or multiple data centers that are geographically separated by metro
distances. Furthermore, with a unique scale-up and scale-out architecture, EMC
VPLEX’s advanced data caching and distributed cache coherency provide workload
resiliency, automatic sharing, and balancing and failover of storage domains, and
enable both local and remote data access with predictable service levels. Oracle
Extended RAC dispersed in two datacenters within metro distance backed by the
capabilities of EMC VPLEX provides improved performance, scalability, and flexibility.
In addition, the capability of EMC VPLEX to provide nondisruptive, heterogeneous
data movement, and volume management functionality within synchronous distances
enables customers to offer nimble and cost-effective cloud services spanning
multiple physical locations.
References
The following documents include more information on VPLEX and can be found on
EMC.com and Powerlink:
Implementation and Planning Best Practices for EMC VPLEX Technical Notes
EMC VPLEX Metro Witness – Technology and High Availability TechBook
Conditions for stretched hosts cluster support on EMC VPLEX Metro
EMC VPLEX with GeoSynchrony 5.0 Product Guide

h8930 Vplex Metro Oracle Rac WP

Uploaded by

Copyright:

Available Formats

h8930 Vplex Metro Oracle Rac 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

h8930 Vplex Metro Oracle Rac WP

Uploaded by

Copyright:

Available Formats

White Paper

Oracle Extended RAC with

EMC believes the information in this publication is accurate of

The information in this publication is provided “as is”. EMC

Use, copying, and distribution of any EMC software described in

VMware is a registered trademark of VMware, Inc. All other

Part Number h8930

Oracle Extended RAC with EMC VPLEX Metro 2

Oracle Extended RAC with EMC VPLEX Metro 3

Oracle Extended RAC with EMC VPLEX Metro 4

Oracle Extended RAC with EMC VPLEX Metro 5

Products and features overview

Figure 1. EMC VPLEX Local to federate heterogeneous storage

Oracle Extended RAC with EMC VPLEX Metro 6

VPLEX product offerings

Figure 2. VPLEX topologies

VPLEX Local allows increased availability, simplified management, and improved

VPLEX Metro with AccessAnywhere

Oracle Extended RAC with EMC VPLEX Metro 7

Figure 3. Oracle VM Live Migration with VPLEX Metro

High Availability Infrastructure — Reduces recovery time objective (RTO). High

Figure 4. Oracle Extended RAC with VPLEX Metro

Oracle Extended RAC with EMC VPLEX Metro 8

VPLEX architecture highlights1

VPLEX Cluster Contains one, two, or four engines

VPLEX Engine Contains two directors, management modules, power

Oracle Extended RAC with EMC VPLEX Metro 9

Figure 5. EMC VPLEX logical storage structures

Oracle Extended RAC with EMC VPLEX Metro 10

There are two types of consistency groups:

Synchronous consistency groups — Provide a convenient way to apply the same

Asynchronous consistency groups — Used for distributed volumes in a VPLEX Geo,

Oracle Extended RAC with EMC VPLEX Metro 11

Asynchronous consistency groups detach rules can be Active Cluster Wins or No

Oracle Extended RAC with EMC VPLEX Metro 12

Oracle Real Application Clusters

Oracle Automatic Storage Management

Oracle RAC on Extended Distance Clusters

Oracle Extended RAC with EMC VPLEX Metro 13

Symmetrix VMAX TimeFinder product family

Oracle Extended RAC with EMC VPLEX Metro 14

Implementing VPLEX Metro with Oracle Extended RAC

Oracle Extended RAC with EMC VPLEX Metro 15

Oracle Clusterware and VPLEX Witness deployment

Oracle Clusterware deployment

Oracle Extended RAC with EMC VPLEX Metro 16

Oracle Extended RAC with EMC VPLEX Metro 17

Resiliency to host and site failure scenarios

Host HBA port Each host should have

Oracle Extended RAC with EMC VPLEX Metro 18

Resiliency to database/network related failure scenarios

Database instance Downtime implied until Oracle RAC provides

RAC Cluster N/A RAC natively and

Resiliency to Storage failure scenarios

Front-end port SAN connectivity should

Physical drive Use storage RAID

Oracle Extended RAC with EMC VPLEX Metro 19

Storage array Symmetrix components are

Resiliency to VPLEX failure scenarios