Replica from backup copy

[lowlander]

Hi,

Can someone clarify on creating a replica based on a backup copy repository ?

step 1) Daily we create a backup of a virtual machine on site A and copy it to site B. On site B there is a repository (XYZ) containing the backup copy. We use WAN accelerators between the sites.

step 2) After the backup and backup copy we initiate a replication of the virtual machine on site A to site B. Within the virtual machine selection we select repository XYZ as the repository that the replication job should be obtaining VM data from. Within the job settings we us a local repository on site A for meta data.

When we monitor bandwidth for step 2 (the replication of the vm), we see a lot of bandwith being used between site A and B. In detail there is a lot of inbound bandwidth on site A.

In this scenario, should we select the meta data repository on site A or site B. Best practices is to select A, but thinking it over the replica is build up based on the backup copy on site B. Is this a possible reason for the large amount of inbound bandwidth we see ?

thanks !

[Vitaliy S.]

Hi,

Metadata should be stored close to the source VM image. In your case it is site B and the repository you're replicating VMs from. Please reconfigure your replication job to use this repository and let us know about network traffic consumption.

Thanks!

[lowlander]

good point, we have tested it : We stored the meta data on site B for the replication job. When we monitor the Veeam Backup server on site B (during replication) we see outgoing network traffic being generated by the image process : Veeam agent.exe. So veeamagent.exe is sending traffic from site B to site A, using all the bandwidth it can get.

Please advise how to throttle this outbound traffic to site A. Bandwidth throttling between site works as configured however not for the veeamagent.exe proces.

thanks !

[veremin]

As long as repository storing replica digests resides in site B, there should not be an excessive traffic going from B to A. If you the opposite behaviour, kindly, open a ticket with our support team and let them analyze that directly. Thanks.

[Vitaliy S.]

Just to make sure that you've done everything correctly before opening a support case - do you see the same amount of traffic being sent to site A when metadata is stored in site A and then in site B?

[lowlander]

Hi turns out to be related to WAN Acceleration and the location of the meta data. We used primary WAN accelerator when not using the backup repository on site B in the past. We changed the WAN accelerator transport mode to direct mode, now we suspect there is no more inbound traffic on site A. I will check the WAN bandwidth usage between the sites, and revert on this.

[Vitaliy S.]

Yes, good catch.