Hello Everyone,
I am hoping that someone here have seen this issue or replicate this issue and possively provide a solution to my "problem".
Here are my senerios:
- BE 2015 FP3 (latest) on Windows 2008R2 Enterprise
- 10GB 9000MTU iSCSI backend into EMC SAN with multipath IO
- VMware ESXi 6u1 hosts
- vCenter applicance 6u1
- Backup transport mode is SAN
- VM's are all located on SAN
- Full VM backup
I've created 2 backup jobs, one for a Linux VM and one for a Windows VM. They are both located on the same datastore and host and have very simlilar disk size.
Here is the problem:
- Linux backup job rate is about 5000-8000MB/m to tape and disk
- Windows backup job rate is 1500-2500MB/m to tape and disk
I've tried various settings and backup themods such as GRT/no GRT, AOF on and off MS VSS on and off, IPv4, latest backup agents and everything that I can think off but still no luck in figuring out why the Windows job is so much slower than the Linux job.
Doing some network monitoring, I've determined that during a the Linux backup job, there isn't much communication going on between the BE server and the SAN. Hower when doing the Windows backup job, there is a constant request from the BE server to the SAN for 512 blocks at time (blocks can be various but mostly 512)
Basically, the only difference between the Windows job and the Linux job is of course the OS. And you wouldn't think that it matters to the BE server since it's backing up the blocks.
Is BE 2015 treating Windows backup job different than Linux?
Any ideas, sugessions or educated guesses?
Thanks,
Ken
