SQL Database restore - network traffic flow?

[yakamoneye18]

Hey Community,

I have a question about the network traffic flow during a database restore:

We have our main site, where we have a sql server, Veeam B&R Enterprise, and the repository. On a secondary location a few km away (connected via MPLS 1G) we have a secondary vmware cluster where we replicate all vms after the backup jobs for worst case scenarios. On this cluster we have a Veeam backup proxy machine that handles the replication traffic (throttling rules enabled).

Now we have new hardware for the replication and have a lot of unused space on the storage - now we wanted to move our test SQL server to this cluster. The only reason why we did not do this until now is because we use Veeam SQL Restore to create new SQL test databases whenever needed: we just "restore" the live backup to the test server. At the moment, we do this with 2-3GB/s because testserver, Veeam server and repository have 10G connections. This takes about 40 minutes. If we do this over the 1G line, which in addition is throttled to 700mbit max, the restore would take much longer.

My question: if I would use the backup proxy in the secondary location - or aother system there - as target for a backup copy of the sql server; I would make incremental copies, so the load would be minimal. Would the restore traffic still go through the main B&R server, or would the data only move between the backup copy repository and the test sql server?

Thanks in advance for picking your brains...

Regards,

Tobias

[Mildur]

Hi Tobias

Restore Traffic for restoring an entire database flows from the assigned mount server of the backup repository to the target MSSQL server.
If you do a Point In Time Restore or fine-tune recovery, restore traffic will flow from the mount server to the staging server and then to the target MSSQL server.

You find the port requirements here:
https://helpcenter.veeam.com/docs/backu ... 10#restore

Have you tried Publishing the database to the MSSQL server? Depending on what you need to test this could be an option.
https://helpcenter.veeam.com/docs/backu ... ml?ver=110

Best,
Fabian

[yakamoneye18]

Hi Fabian,

thanks for your reply.

Publishing is not an option in our case - we permanently have some test databases, where developers and our users can test processes in our ERP - form time to time we refresh the databases with a current data state.

I checked the other link you sent - so this means if I make the proxy in the remote site a repository and assign itself as mount server, and the database server is located on the same esx, the restore process should use the full available network speed of esx-internal network, even if I start the restore process from the console on the main site?

[Mildur]

Hi Tobias

Yes.
Your SQL Server will mount the VM filesystem from the restore point over iSCSI from the assigned mount server.
https://helpcenter.veeam.com/docs/backu ... ml?ver=110

After the mount operation, the Veeam SQL Restore Service takes over. It runs on the target MSSQL server.
It copies the database and log files over from the mounted VM filesystem.
https://helpcenter.veeam.com/docs/backu ... ml?ver=110

Your Veeam Console needs to have access to the MSSCL Server and Veeam components for orchestrating the restore process. But the main data flow is as I have explained.

Best,
Fabian