Very slow Hyper-V Snapshot merge?

I use latest Veeam Backup and Replication to create replication task between my 2 Hyper-V Host, the first Host is production (live) and second host is a failover replica (offline). Actually Veeam is only "scripting" the built-in feature out of
my Windows Server 2012 R2 Hyper-V to create backup and snapshots between my two hosts. Everything is working fine except for one thing.

When Veeam launch the task to merge the oldest VM snapshot (I keep 3x snapshots) on the Replica Host, the "merging" process is taking A WHILE to complete. I NEVER FAIL, but it take A WHILE, like 2+ hours!

I can’t identify why the "merge" is taking so long. My replica Hyper-V Host is a VERY high performance HP Proliant Server, with a lot of RAM, lot of CPU, lot of everything. The storage for the server is a very fast QNAP Entreprise NAS, connected
to the server using iSCSI. If I run read and write benchmark to my QNAP NAS, it’s very fast. Since the NAS is connected with Gigabit ethernet for the moment (don’t have 10 Gbps on my switches, but the QNAP have 10 Gpbs), I have a super stable 110 MB/s in both
read and write, which is so far, excellent!

The "affected" VM of the slow merge is my main file server VM, who have obviously large VHDX volumes. The main VHDX volume is 6.5 TB but snapshots are only around 100-125 GB but the "merging" process is taking more or less 2 HOURS to
complete. The VHDX volume are pre-allocated if this matters (as in the best practice in Hyper-V for production drive).

BY THE WAY, I already opened a service ticket with Veeam and they have answered that the Hyper-V snapshot merging is a Microsoft feature that is only initiated by Veeam, so that they can’t determine why the process is taking so long because it’s on the Hyper-V
Host side.

So if my storage and my server is fast, why did it take that long to merge the Hyper-V Snapshot???

Read full post . . . .

Go Que Newsroom