Excluding VMs from Datastore-based backup

[Nevin]

Hi All,

I'm trying to design our Veeam backup solution and our environment is not exactly "ideal". I'm hoping someone whose in the same boat will have some insight.

80% of our VMs are on direct-attached storage to the 5 ESXi hosts we have. The rest are on a SAN and yes we do have a vCenter Server.

My plan is to install a Veeam Backup & Replication Virtual Appliance on every ESXi host, and then use the Enterprise Manager as well.

Here's my problem. When I'm creating a backup job if I pick and choose every VM that needs to be backed up, what happens when one moves? Am I going to have to manually remove the VM from that job and then add it to the job on the server the VM now resides on? The hosts cannot see each others arrays, so I can't think of how else to do this.

I considered doing a datastore-based job, but that didn't work out either because I couldn't figure out how to exclude certain VMs from the jobs. We have a lot of sandbox-type VMs that do not need to be backed up in these jobs.

Anyone have any ideas?

Thanks,
Nevin

[Vitaliy S.]

Hello Nevin,

Well... in order to continue using Virtual Appliance mode only with local storage, you would need to reconfigure your jobs. But could you tell me how often will you have VM migration between hosts' local storages? You can configure each job through a vCenter Server and use failover to network mode option when VA mode is not available.

In other words, when you add your VMs through a vCenter Server connection Veeam backup server will still be able to reach these VMs by a unique moref ID generated by vCenter Server. Even if this VM will be moved to a different local storage, Veeam backup server will backup this VM via network mode automatically. Besides, with VMware Changed Block Tracking enabled you won't see a signifficant difference in VM processing rates on incremetal job run.

Hope it helps!

[Nevin]

Hello Nevin,

Well... in order to continue using Virtual Appliance mode only with local storage, you would need to reconfigure your jobs. But could you tell me how often will you have VM migration between hosts' local storages? You can configure each job through a vCenter Server and use failover to network mode option when VA mode is not available.

More than we should, that's for sure! The VMs are slowly being rearranged right now. After that the movement will be minimal.

Besides, with VMware Changed Block Tracking enabled you won't see a signifficant difference in VM processing rates on incremetal job run.

Are you saying that there won't be much of a benefit from using HOTADD over NBD as long as CBT is enabled? If that's the case maybe I won't make so many VMs just to use virtual appliance mode.

Hope it helps!

Very much! Thanks for your time!

[Vitaliy S.]

Are you saying that there won't be much of a benefit from using HOTADD over NBD as long as CBT is enabled? If that's the case maybe I won't make so many VMs just to use virtual appliance mode.

Not really, of course SAN and VA mode is much faster than Network mode. But what I am trying to say is that for VMs with minimal changes between backup cycles, you'll have a pretty fast incremental run even with Network mode, meaning that you should be able to fit in your backup window. By the way have you already tried running any test job?

[Nevin]

Sorry for the delay there. Was getting all our hosts to 4.1 this weekend.

Yes, I have run some tests and they do seem quite fast. I didn't really do much of anything in NBD mode until now.

I do see the process rates vary a lot, anywhere between 16-320 MB/s . Probably averages out in the 100s somewhere. I think the fastest network mode was 120MB/s.

I can't say I entirely trust the rate, though. I think it is accounting for unallocated space. If I've got a 500GB disk that's thin provisioned and it's just got a vanilla XP install, is that rate going to take into account that it backed up 490GB of nothing? I don't think it is but I could be wrong. Either way, it is faster than the Fulls we do with Backup Exec.

We should be purchasing enough licensing soon so that we can use this as our primary backup solution. Right now I'm focusing on setting this up so there is as little user intervention as possible. We are absolutely sick of micro-managing Backup Exec and that's one of the reasons we wanted a different solution.

[Gostev]

I considered doing a datastore-based job, but that didn't work out either because I couldn't figure out how to exclude certain VMs from the jobs. We have a lot of sandbox-type VMs that do not need to be backed up in these jobs.

Hi Nevin, I do not see this question answered above, so I figured I bump this... just wanted to make sure that you were able to locate the UI option that allows you to exclude individual VMs as well as containers (VM folders, resource pools, etc.) from the backup job. We do have this capability in the Exclusions dialog. Thanks!