2 VMware Clusters

[jcwuerfl]

If you have 2 VMware clusters and both clusters have separate non-shared Storage between clusters what's the best way to deploy Veeam as far as Backup Proxies?

ClusterA - Datastore1
- VM 1 with Backup Proxy #1

ClusterB - Datastore2
- VM 2 with Backup Proxy # 2

Thinking that otherwise it wouldn't be able to do the Hot-Add feature if you don't have a VM that has access to Datastore #2 ?

[jcwuerfl]

Another example is what if you have say 3 standalone licensed vmware ESXi servers with just local disk?

ESXi1 - Local Disk
- VM 1 - with Veeam Backup Proxy #1

ESXi2 - Local Disk
- VM 2 - with Veeam Backup Proxy #2

ESXi3 - Local Disk
- VM 3 - with Veeam Backup Proxy #3

[bmeyer99]

All your scenarios are valid and will work. You can also deploy a single physical proxy server and mount all the datastores from your two clusters to it and then use Direct SAN Access. In the standalone hosts, how you have it laid out is your best bet.

[jcwuerfl]

Going back to the cluster A | B if there are 6 hosts in cluster A and 3 hosts in cluster B does there need to be a backup proxy on each host? or would 1 per cluster be ok?

[dellock6]

It depends on the speed you can reach extracting data from your san, and how fast your repository can accept backup data. The bare minimum, if you are going for virtual appliance mode, is a single virtual proxy hosted in a ESXi server connected to all the datastore you need to backup from. From here on, you add more proxies "only" for speeding up the backups.

[bmeyer99]

@dellock6 is correct. @jcwuerfl if you have 1 proxy in Cluster A and 1 proxy in Cluster B and you are backing up all your VMs within your backup window then I wouldn't worry about adding any more proxies. If you would like to speed up your backups by running more jobs simultaneously then add another proxy to Cluster A. However since all proxies usually go to the same Repository you need to make sure you repository is sized correctly to accept that many data feeds, as well as configured within Veeam to accept the number of multiple feeds or ingestion rate. Utilize the Bottleneck indicator within the job history to see where your bottleneck is. If you are seeing low percentage on the Target then your repository is keeping up just fine.

[jcwuerfl]

Ok, thanks all! Was just curious as I haven't set that up yet on if you had to do it per host or if it was more just having access to the datastores. So glad its the latter really unless you need more performance.

Thanks!