HBR Disks - How Big Do I Plan For

[sutcliff]

I am pretty new to SRM. I have been using replication for a little bit but now have run out of space. I was not planning for the HBR disk space. From what I have googled it appears they are either redo logs or snapshots.

How much space should I plan for HBR files? I see that one of my replicated disks has two  HBR disks. One is 16GB and the other is 1MB. The other virtual disk has a 19GB HBR file that blew out my space.

Does the size growth depend on how much has changed?

Thanks,

Brian

[mvalkanov]

Hi Brian,

vSphere Replication keeps delta disks until consistent replica instance is received. Then it consolidates the replica instance with the base disks at the target datastore. The delta disks grow with all changed blocks. A changed disk block is replicated only once, even if it has changed multiple times within one RPO window.

At worst case, within a single RPO window, the delta disks can become as large as the base disks.

In addition, if you perform Test Failover, the delta disk hierarchy is forked - incoming replica data does not interfere with the test VM. Once Test Cleanup is executed, the hierarchy is consolidated.

You need to plan for worst case 3x space of the source VM - 1x for the latest consistent instance, 1x max for incoming changes and 1x max for changes done when test VM is powered on. In most of the time, there won't be test VM, so 1x + amount of changes is the space occupied.

If you are using VR 5.5 and have enabled MPIT (Multiple Point In Time Recovery), you need to plan for each MPIT instance - each instance will contain delta for the changes from the previous instance.

Regards,

Martin