Habeeb,

 

We've said all along that DR for VCS MC v5.1 would not make use of GCO or data replication.  The peering solution would be the DR solution.  The only other decision then is whether to make the management server itself highly available.

 

The reasons for not using GCO and replication involves how VCS MC manages it's data.  Remember that a management server will only handle reporting for clusters at the local site, so even if the data was replicated the management server at the DR wouldn't know how to deal with the other site's reporting data.  It has no mechanism for handling reporting for servers at a remote site, other than the use of another management server.  With peering between sites, both management servers can manage the entire enterprise, so long as both of them are up - thus the need for HA.  If either management server goes down, then the surviving management server cannot manage the clusters that were registered by the failed management server.

 

Let's talk about failure modes where we have 2 sites, each running a management server.  If we loose the IP connectivity between the management servers, they still can manage the clusters and reporting at each site - there will be no peering however.

 

If we lose site A completely, then the site B management server just has to worry about site B's clusters - no problem here.  If we lose part of site A where the management server and some clusters still survive - no problem.  We have the management server up and running to manage the local clusters.  If the IP link has survived the disaster then peering will happen, if not peering will not happen.

 

So the short of all this is if you have a management server at your production and your DR site, there is no real reason to either replicate management server data, or use GCO for DR.  The VCS MC architecture is inherently disaster tolerant if peering is used.

 

Regards,

Hugh