SRM placeholder datastore design

By | January 26, 2015

Recently I was doing a VMware project with 14 HA Clusters (DC) and 8 HA Clusters (DRC) . During the design considerations I wondered how many datastores should be created for Site Recovery Manager (SRM)  placeholders. The placeholder is a file which holds configuration information corresponding to the Protected virtual machines.

Following VMware documentation placeholder datastores must meet certain criteria:

  • for clusters, the placeholder datastores must be visible to all of the hosts in the cluster.
  • you cannot select replicated datastores as placeholder datastores.

When we design placeholder datastores, generally we have 3 options:

  • dedicated placeholder datastore for cluster HA.
  • shared placeholder datastore across clusters HA.
  • 2 SRM placeholders datastore: 1 for each site.

If you have only one cluster HA per site, the answer is simple: one placeholder datastore for site.  So what if you have more clusters? It depends on (e.g. size of infrastructure) - there is no technical limitation to share datastores across clusters so  I create shared placeholder datastore because it is easier for administering. Sometimes creating shared datastores is limited e.g. the project I mentioned at the beginning of this article - I have a customer where sharing datastore across cluster is prohibited (security regulations).

Note: I have met SRM configuration with placeholders located at the same datastore with VM files! Of course it is NOT RECOMMENDED but also possible configuration...

 

3 thoughts on “SRM placeholder datastore design

  1. Javier Martinez

    How can I know the correct size for the Placeholder Datastore???

    Reply
    1. Mariusz Post author

      The easiest way is around 1MB (in practice much less) for each protected VM. 1GB datastore is enough in almost all cases.

      Reply

Leave a Reply to Javier Martinez Cancel reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.