Full VM Restore issue

[stack111]

Hi,

Our customer Veeam backup and replication 9.5 update 3 server is connected with EMC data domain via fiber but when we try to restore a VM with 120 GB of size from EMC data domain to the original location on the Hyper-V cluster, it took for the Veeam server around 30 minutes to finish the restore which is not logic as the connectivity is fiber, we need to know why it took that amount of time and fix any issue if exists.

I already logged a ticket on Veeam portal, ticket ID is 02687240

[stack111]

Hi,

Veeam support replies as below:

The EMC Datadomain is a great storage device. Because of the deduplication you can store an incredible amount of data. Because of the deduplication the restore process from an EMC datadomain will be significantly slower. Here is a best practices KB for EMC Datadomain. As a first step please review.

https://www.veeam.com/kb1956

As a second step can you send us the logs so we can see where the bottleneck of your restore is occurring. I assume is it a full restore. If so the logs for the restore will be located in C:\programdata\veeam\backup\"Server Name"

I sent already the logs and I am waiting their feedback

[stack111]

This is support reply

Thank you for logs. As far as I can see, the main bottleneck is read from the source (from datadomain).

Agent.servername.Restore.Source.log
99/6/0

Restore is a technology that heavily depends on random Read/Write I/O, meaning that if a storage device can not provide that appropriate level of I/O the technology will have trouble performing as expected. There is no "fix" that Veeam can provide to speed up the datadomain's random read/write I/O to make the restore run faster. In Veeam 9.5 update 3 [https://www.veeam.com/kb2353] there were some improvements to the performance of the restore and the VeeamNFS service, but I doubt one would notice any difference if the restore points are located on a deduped storage.

[stack111]

Hi,

I have a question related to Veeam Backup and Replication 9.5 update 3 restore mechanism, when I restore a full VM from EMC data domain connected to Veeam server via Fiber channel to the original location on the Hyper-V cluster, does the restore to the destination happens via fiber (Hyper-V servers are connected to storage via fiber) or it is over network?

[foggy]

Hi Ahmad, as far as I an see, support engineer has already explained you the reasons of the observed restore speed (not connectivity but rather read from dedupe device). Do you have any additional questions?

[stack111]

Actually this is a new question but not sure why it is merged here...I already got support response and I understand it

Because we have been asked by our customer if the restore to the Hyper-V cluster is done via network or fiber