Users Guide

Figure 4. MD Storage Manager Snapshot Virtual Disk View

How SRA uses snapshots

The MD SRA uses Point-in-Time Snapshots if the feature is enabled on the storage array. During test failover, the SRA creates a

Snapshot Group, Snapshot Image, and Snapshot Virtual Disk on the recovery site’s storage array for all virtual disks contained in the

protection groups being tested. This process requires the creation of the two snapshot repositories, listed earlier. The default size for

these repositories is 10 percent of the base virtual disks for each repository, for a total of 20 percent of the base virtual disk size,

which means that the free disk space on the recovery site storage array must be 20 percent of the base virtual disks participating in

the test failover virtual disks. This value is controlled by the SraCongurationData.xml le located in the cong directory under the

installation directory, typically:

C:\Program Files\VMware\VMware vCenter Site Recovery Manager\storage\sra\MD\cong\SraCongurationData.xml

The value is set with the XML tag <SnapshotBasePercentage>.

<!--

SnapshotBasePercentage represents the initial size, expressed as a percentage of virtual

disk size, of a snapshot which is formed for test failover.

-->

On the basis of how you use the VMs during test failover, you can ne-tune this value for your environment to reduce the free disk

space required for test failover. If the test VM that is stored on the recovery site ESX host does not write extensive data to the

Datastores and no synchronization (or minimal changes) occur between the protected site virtual disks and the recovery site virtual

disks during test failover, you may decrease this value to 2–5 for even less free disk space during test failover. The Snapshot Virtual

Disks and Snapshot Group repositories are deleted during the cleanup phase of test failover along with the Snapshot Image.

Because the Snapshot Virtual Disk is typically not used during the test failover process (minimal write activity), you may decrease

the size of these repositories to small sizes in order to preserve free disk space on the recovery site’s storage array. If a repository

runs out of disk space during the test failover phase, the VMs on the recovery site lose access to the Datastore and underlying

virtual disk is aected by the out-of-space condition of the repository, but the protected site VM functions as normal. Dell does not