On Fri, Feb 5, 2016 at 1:41 PM, Charles F Sullivan <[email protected]> wrote: > > Is there any reason I should be afraid to use VMware replication to make > copies of our DCs in the event of a data center-wide disaster?
I use SRM (Site Recovery manager) to replicate 4 different DCs. 1 Forest, 2 domains. In my case, however, we don't use VMware replication, we use SAN replication via EMC's RecoverPoint appliances. > The replication would be set with an RPO of 15 minutes. In a disaster > scenario for our data center, the DC at the other site would be the only one > standing, but I would bring up the replicated DCs, one at a time, starting > with the PDCe. The only other thing I would need would be to confirm that the > IP configuration holds or set it correctly if needed. I have physical DCs at remote sites (we have 4 sites), so I'm covered that way. But during our last DR test, we did successfully fail over all the VMs using SRM, and all the VMs did start. I'm pretty sure that the DC VMs just started replicating, once they came back online over at the DR site (it was like a year or so ago). And yes, the PDCe is one of the VMs. In our case, the IP configuration holds, so basically it was like everything was still running at the main datacenter. (I'm not involved in the networking side - i.e., the configuration of the physical switches, etc, so I don't know what those guys had to do. I do know that we set 2 gateways entries, one for the main datacenter switch, one pointing at the switch at the recovery site, so if it doesn't find one, it finds the other. It took a while, because most of the VMs are set to *not* autostart. I manually start the VMs via a script (based on what folder the VM is in; we have a folder for each DR priority - Critical 1, Critical 2, etc). I don't want to flood the reduced bandwidth with a bunch of VMs, all looking for a DC at the same time, as well as other connections they might be looking for (database connections, etc) HTH
