Hello,
I have an SRM installation with 2 sites. The sites are seperated by metro ethernet layer 2 trunk between sites.
Site A
1 EqualLogic PS6100
2 ESX servers configured as a cluster
There is a single datastore volume on the SAN holding all VMs in this cluster
There is one instance of vCenter at this site that resides on the same datastore as the other VMs
There are also a few other volumes on the SAN used for RDM disks within the VMs
I have also created a small 1GB volume as a VMFS datastore and have it mapped within SRM to use as a placeholder datastore
What are your thoughts on this? Is this the best way? Could I use local storage on the ESX servers? Or can my main VM datastore be used for placeholders?
SRM is installed on the same VM as vCenter
SRM is using SQL Express on the local vCenter VM
vCenter is also using SQL Express
Site B
1 EqualLogic PS4100
1 ESX server, so no clustering
There is a single datastore volume on the SAN holding all VMs running on the ESX server
Per SRM requirements, there is a seperate vCenter instance running as a VM, and it is not in linked mode. Should it be? / Does it need to be?
The number of volumes on the SAN and their purposes are the same as in Site A, only difference is the volume sizes are smaller because less VMs running at this site.
Replication has been set up and verified working bi-directional.
We are using array-based replication
Ok, so here is the problem... When I do a planned recovery, the process shuts down all VMs at that site which includes the vCenter/SRM server. I assume this happens because it is on the same datastore as my other VMs?
If I run recovery in disaster mode, then vCenter is moved to the other site and powered on. Because sites are layer 2 and on same vlan, SRM shows the sites are connected due to the 2 SRM instances can see each other eventhough they are now running at the same site. This does not seem correct to me.
Can anyone suggest a better configuration?
Thanks,
Brian