Advice about large Server

[ober72]

Hi Folks,

I have a very large exchange (7 drives, a few are multi terabyte). When we first tried to back this up with Veeam occasionally the VM would get stunned for a number of minutes causing outages and the snapshot removal took a very very long time. In the end desperate we installed Endpoint which works but takes 12 hours to backup as the network for now is only 1Gbs.

I want to try using virtual appliance transport mode with the vmware hot-add feature but I need to know how risky this is for the Esxi, i.e. will this cause another outage. I believe that I could install the veeamone and use the backup assessment tool to get some information on the infrastructure but is there anything else that I could do beforehand to see if this is worth the effort?

I am hoping that hot-add would significantly improve performance although I worry that the hot add process puts extra load on the hyper visor and it will still take a while to remove the vmware snapshot.

thanks

[Shestakov]

Hi Geoff,
What is usually a bottleneck for your backup jobs of VMs running on the same ESXi and using same proxy and repository?
You may also be interested in the related thread discussing backups of large VMs.
Thanks!

[ober72]

Hi Nikita,

The bottleneck is always 99% 98% source . There are a bunch of ESXi hosts in the cluster, using the default proxy on the Veeam appliance jobs go to two different repositories. On a side note I don't have any access to the Esxs other than what I can see in the VBR console.

Perhaps adding a proxy in a vm on the management network would be safer than virtual appliance mode but still improve performance?

The people that i am dealing with are very allergic to any down time whatsoever even a few minutes as their monitoring software sends out email alerts.

thanks

[Shestakov]

Having source as the bottleneck, it`s recommended to leverage hot-add or direct access transport mode.
I would give it a try on smaller VMs first to see how it works if you want to be cautious.
By the way, do you use Veeam ONE for monitoring?
Thanks!

[ober72]

Not here but it may be an option. In fact if I am not mistaken Veeam ONE free can monitor the hyper-visors?

Thanks I will try the hot add on a small vm.

cheers

[Shestakov]

Yes, virtual infrastructure monitoring is included in free version of Veeam ONE.

The people that i am dealing with are very allergic to any down time whatsoever even a few minutes as their monitoring software sends out email alerts.

If you use Veeam ONE for backup infrastructure monitoring, you can suppress particular alerts for the time of backup activity.

[skrause]

If you are trying to avoid the snapshot-removal stun, none of the "standard" Veeam B&R methods (Hot Add/Network/Direct SAN) will be able to avoid this. The only method that potentially will give you a stun-free backup is Backup from Storage Snapshots which requires having primary storage that supports this feature with Veeam.

Snapshot removal stuns are pretty common on servers with many/large VMDK files, especially ones that have a high change rate. This is a "feature" of VMware itself.

[ober72]

Ah!! that is kind of what I thought but was naively hoping that it was not the case. Hence why they re-engineered the whole CBT on Vmware 6, at least I think I saw that somewhere. Any best practices for minimizing the stun again wishful thinking. My assumption is that the bigger the disks and the more of them the bigger potential stun.

[foggy]

Actually, the more active the VM is, the larger is the snapshot and, hence, longer commit. Some hints can be found in this huge thread.