HP StorageWorks Continuous Access EVA Administrator Guide (T3687-96043, December 2005)

risk. If a major failure occurs at the local site during a full copy, the snapclone provides a clean copy
of data as it existed before full copy writes started on the remote array. (Any new writes that occurred
on the source bet ween the time the snapclone was created and the m a jor failure occurred would b e
lost.) As a best practice, whenever a link is expected to be down more than several minutes, create a
snapclone of the destination virtual disk.
NOTE:
You cannot use this procedure if a full copy has been started.
1. Using the replication manager, navigate to e ach affected DR group and suspend replication.
2. Make a snapclone of the destination virtual disks, using the procedures described in HP
StorageWorks Replication Solutions Manager online help and user guide.
3. Using the replication manager, navigate to each affected DR group and resume replication.
Replication will resume once links are restored.
Optimizing performance
Load balancing
For best performance of remote replication, the average workload (reads and writes) should be equal
on both controllers, and on both fabrics and interswitch links. To o btain this balance, take manual
measurements and keep the utilization rate of each interswitch link b elow 40 percent. If one link fails, the
average utilization of the surviving link would n ot exceed 80 percent. Similarly, the utilization rate of a
single controller as measured on all host ports should not exceed 45 percent on average, or peak above
50 percent, to prevent overloading a surviving controller should one of them fail.
In general, let hosts manage load balancing. To support host load balancing, all members of a DR group
must share a preferre d controller path and HBA por t pair. Load balancing a single application with all
its virtual disks in a single D R group (as required) is not possible.
Minimize simultaneous replication events
Minimize the number of replication requests to the same array at the same time. Consider limiting
access to the various management and command line interfaces. Too many simultaneous replication
events can reduce array performance.
Also, avoid making multiple replication requests to the same virtual disk at the same time. Multiple
replication events to the same virtual disk not only slow performance, but in the case of automated jobs,
can lead to job failures. For example, if the maximum number of snapshots per virtual disk is exce eded
when the job is running, the job will fail.
Backing up replication configuration
HPrecommendsregularbackupsofreplicationconfigurations and jobs. Current backups are essential to
successful disaster recovery. You can also use backups to duplicate the local configuration and jobs on
remote and standby management servers and keep those servers current.
The replication manager and Storage System Scripting Utility (SSSU) provide backup methods, or you can
back up your configuration manually.
Using the replication manager for backups
The replication manag er's export feature copies the replication da tabase, including DR groups, jobs,
and the encrypted HP Command View EVA (Storage Agent) password. You can import the export file to
restore the database or to duplicate the database on another management server. See HP StorageWorks
Replication Solutions Manager administrator guide for export and import procedures.
54
Managing remote replication