We are planning the future installation of OM 2012 in our 2 main datacenters. 1. We use a single management group across both datacenters for DR and operational purposes. 2. We are looking for datacenter A to be our primary and contain the active DBs with 3 management servers there, and 3 in the datacenter B. 3. Server agents in datacenter A will connect to A and servers in datacenter B will connect to management servers in B. 4. A mirrored database server will exist in datacenter B for manual failover in a DR situation (using CNAMEs for the instances). How should we handle Resource Pools in this scenario? Should there be two separate pools (in addition to the automatic ones) or one spanning both datacenters? If we use two separate pools, is there an effective way to direct user console sessions to connect to either resource pool's management servers (for example DNS round robin)?

the management servers should be on fast connections to the DB preferable on the same vlan. imho your are better of with all agents going to 1 datacenter and have a failover (db) ready in the other. But to the question, i think round robin will work, i'm planning on testing this myself but i haven't come round to it. The current ms solution however is to create a nlb cluster of the management servers (the default unicast is very bad for your switches performance and for multicast mode the switches need to be configured), i suppose you could create 2 clusters, 1 in each datacenter. Rob Korving http://jama00.wordpress.com/