We currently have 2 Hyper-V host (Windows Server 2016).

One of them contains all of the running machines, while the other contains the same machines, but as powered off replicas.

To keep them in sync, we are using a replication job in VEEAM Backup & Replication, but the problem we are currently experiencing is that the virtual machines can't access their hard drive for about 5 seconds every time VEEAM runs the replication job.

After some investigation, we discovered that it probably isn't VEEAM that causes the issues, but Hyper-V.

We determined this by manually creating and merging a checkpoint (just like VEEAM does every time it replicates) when VEEAM was turned off.

Is this normal behaviour, or can it be fixed?

  • How big is the snapshot? How many IOPS can the storage do? You need a vast amount of storage IOPS and bandwidth to merge with no performance hiccup. – John Mahowald Nov 19 '19 at 16:04
  • The snapshots are only around 15-25 MB in size, our storage can do around 250 IOPS – kasperB Nov 20 '19 at 9:40

Merging a delta disk like this is I/O intensive. Read source volumes, read checkpoint, write to target volume. I don't know the exact block size, but it could easily be doing a thousand random I/Os and taking seconds on that storage.

Experiment on faster storage that does many more IOPS. Also store checkpoint and replication target disks to be on a storage system independent from the source volume if possible. Likely you can reduce the performance hit, but not get it to zero.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.