Howto backup voip server without causing it to failover

[kjstech]

Hello,

I also have posted this with our phone vendor, but this involves Veeam so I will post here as well.

We use a VOIP phone system from Interactive Intellegence. Its fully virtualized, sans a few hardware elements to get T1's and pots "on the network".

Seems during the backup window, this voip server does what is called a switchover event. This fails over to another server in a DR site. Normally end users would have no indication that this happens. True while most don't, a few random Polycom VVX phones fail to register and support gets calls regarding phones not working. So end users have to reboot those phones. Also the call center folks have to reset their custom views in their phone system client. Anyway thats all for the Phone system vendor...

My question is, is there a way to configure Veeam or VMWare to be "less disruptive" to the guest OS? I notice in the logs for this guest I see "Media SErver is unable to connect to API hosted on the Interaction Center server". "Server proxy lost connection with a Media Server. This usually means the server has been shut down. If the Media Server server has not been shut down intentionally, it may have crashed or a network problem may have caused it to become unreachable. " Also a lot of services seem to crash for example "MRCP SUBSYSTEM has terminated unexpectedly.".

Now we backup everything with Veeam at our HQ, and nothing else seems to be disrupted by veeam. Sure its after hours and sure if you were to run a constant ping -t on a vm during the end of the backup, when disk consolidation occurs you might see some elivated pings and a packet or two dropped here or there... but its usually a very brief interruption.

I think the problem is two fold. One is mostly with the phone system vendor. They need to "relax" their switchover/failover timers so they can allow and sustain a few seconds of poor performance while the snapshot consolidates. Second is with Veeam. Hopefully there is a way to get better performance to minimize that guest os disruption during a backup.

ESXi 5.0, vSphere 5.0 Enterprise Plus
QLogic QLE3242 CNA, firmware 4.16.34
Brocade TurboIron Ti24x 10gbe switches
EMC VNX5200 Unified shared storage, 10gbe to 3 NFS filesystems, each on their own subnet and vmkernel.
SIOC enabled
dV Switches utilized
Veeam running on Server 2008 R2 fully patched guest OS.
Phone system running on Server 2008 R2 fully patched guest OS.
Backing up to FreeNAS device (7.1 TB available).

[veremin]

If the problem occurs mostly during snapshot commit process, there is little we can do, as we rely solely on VMware technology, playing only "requestor" role. Probably, phone vendor is aware of some tweaks or hack that can make the application less sensitive. Thanks.

[Vitaliy S.]

Also placing this VM on the datastore with less load can make snapshot commit operation less painful.