Best Design Scenario

[mwpreston]

Hello Veeamers!!!

Looking for a little bit of advice on a design for our remote locations. Basically we have 1 ESXi host with a few VMs that we would like to possible backup/replicate. Onsite we have another Windows Server box (physical) which will be hosting the backup/replication space via NFS. So my questions for you all is this.

Is it best to install Veeam on the physical box and have it place the replica's on the nfs datastore?

Or, would I be better off to have another VM running on the ESXi box to do the Veeaming?

Or, should I install Veeam on the physical box and have a proxy installed on the ESXi server...?

Obviously we will do our own testing but was just wondering if anyone out there had already done so and might be able to shed a little light..

Thanks,
Mike

[Vitaliy S.]

Hello Mike,

Well, you definitely want to deploy Veeam proxy servers at both sites, so it's option #3. I would suggest using any VM as a proxy. It will allow you to configure replication/backup jobs to run in the HotAdd mode, which, in turn, should give you a pretty decent job performance rate.

Thank you!

[Astirian]

Hi guys,

Sorry to resurrect this but I'm currently in a similar situation. We had a physical Veeam server doing all the work in Network mode and I've since added a VM Proxy which yielded significant performance increase in Virtual Appliance mode! At least 2x times faster for my test VM. I now have 66 VMs finishing at 4am rather than 7am.

One question I do have though is am I even better off forcing every job (we have 12) to use the VM proxy or leaving it to auto-select, or is this simply something I'll just have to test? AFAIK if the VM is to busy, the job will automatically go to the physical box...

Also, I would love to plug the physical box straight into storage but we're running NFS so I'm not sure plugging fibre channel (or even iSCSI) from the physical server into the storage would actually accomplish anything. I'm a storage newb but my understanding is that NFS uses the IP stack over ethernet (we have 1Gb conn) as does iSCSI and FCoE, albeit in blocks.

So if I configured FC from the physical box into the SAN it wouldn't work because the NFS shares aren't configured to serve LUNs to a potential FC (or iSCSI) connection... Is that correct-ish?

Thanks,
- Daragh

[Gostev]

Hi Dragh,

Auto-select for sure, so that your jobs do not stop working when the proxies they are forced to are down or unavailable.
Direct SAN access mode is only supported for block storage (please see sticky FAQ topic).

Thanks!