Hi Ajay;
VDR is really picky about connectivty to the CIFS host! Therefore make sure it (or they) dont run any other services, A/V SW or anything else that can disrupt the network traffic. Have you got good bandwidth between the ESX- and CIFS hosts; not any qwerky firewalls, overloaded switches? Can your CIFS host deliver the requested I/O? Without being an expert, I can se that several hundred files (slabs) are open for read/write ops at any given time and by nature, I would expect that the more packed the deduped CIFS store becomes, the more critical I/O becomes.
You write that one of the VM's crashed; was this one of your production VM or one of the VDR aps.? It doesnt make sense that your production VM crsh, unless they run out of space on the VMFS store where the VMs snapshot data is being stored?
How large is your CIFS store and how much space have you got free. In my experience it is important to have plenty of free space on the CIFS store; go with at least 20-25% overcapacity. A complicating factor here is, that depending on the numer of VMs, the type of VM and the retension policy, this can be very difficult to plan for, so if possible, max it to 500GB if you can and try not to populate it above 400GB.
Hope this helps.
Regards
Thomas
