Brocade Communications Systems StorageWorks 4400 - Enterprise Virtual Array User Manual download pdf (Page 35)

Storage administration

 EVA disk group layout: To optimize performance and provide fault tolerance, place application

databases files, transaction logs, rich media content, and ESX VM files on separate physical disks in

different disk groups. This enables recovery of data to the point of failure. If the log files and databases

are maintained on the same disk group, in the unlikely event of a disk group failure, recovery to the point

of the last known good backup is the only possibility. Performance also improves because database I/O

operations are random, whereas transaction log and rich media file operations are sequential. Having

both on the same physical disk can cause I/O performance degradation.

 Implicit LUN transition is disabled for EVA Vdisks configured for data replication. HP recommends

balancing the Continuous Access DR group across EVA controllers to maintain an equal workload and to

use concurrent replication tunnels. Set the Vdisks PREFRRED_PATH by using Command View EVA.

Multipath settings on the ESX host must be set accordingly to avoid excessive proxy reads.

 To simplify and automate remote replication link bandwidth sizing, use HP StorageWorks Essentials Data

Replication Designer. Data Replication Designer (DRD) is a free GUI-based tool that simplifies collection

of application profile data from customer application servers and automatically calculates the size of the

link required for remote replication that meets the business requirements. This tool is available for

approved HP partners. For more information about Data Replication Designer, consult your HP

representative or Storage Architect.

Implementation caveats

Renamed Vdisks on the EVA

The EVA SRA requires the source and destination Vdisk to have the same name. Command View EVA

allows renaming of either the source or destination Vdisk in a copy set. However, if the virtual disk name is

different on source and destination array, Site Recovery Manager will not correctly identify the protected

disk and will fail to create the snapshot when executing a recovery plan in test mode.

HBA rescan problem

In our testing, we sometimes observed that QLogic HBAs needed to be rescanned twice before the EVA

LUNs (VMFS volumes or raw-mapped LUNs) are discovered. When using Site Recovery Manager, the HBA

rescan only occurs once by default. The default Site Recovery Manager rescan behavior can be changed in

the vmware-dr.xml file located in the c:\Program Files\VMware\VMware Site Recovery Manager\config

directory.

For more information, see the VMware KB 1008283.

Protection group to DR group relationship

There is no mechanism in Site Recovery Manager that prevents an administrator from creating two different

protection groups that use Vdisks contained within the same Continuous Access DR group. Configuring a

protection group this way could have dire consequences.

1 2 ... 30 31 32 33 34 35 36 37 38

No comments

Brocade Communications Systems StorageWorks 4400 - Enterprise Virtual Array User Manual Page 35