SRM replication and Veeam Backups causing issues

[saintdle]

I have a customer whom is using SRM to replicate critical VM's throughout the day (every 1 and 2 hours).

Each night Veeam (version 7, updating to version 8 today) backs the VM's up to a NAS device.

The production VMware environment is using NetApp SAN over FC, a synology NAS is used as the backup repo. A virtual machine is selected to be veeam proxy.

The DR environment is geographically near, here there is another set of hosts with NetApp SAN over FC, the physical Veeam server is also located here.

When backup jobs run, it uses the veeam proxy in the production environment, jobs show as HOT ADD.

it's become apparent that there is an issue that SRM is replicating machines whilst VMs are being backed up by Veeam, meaning there can be two open snapshots per machine, sometimes the snapshots are not removed properly, which leads to performance issues.

During the actual backups, servers such as exchange may "go offline" for a period of up to 30 minutes.

I've looked into the latest version of SRM, but it seems you are not able to set a replication job window, so that replication only happens during business hours. Due to the setup of the company, they want to remain using SRM and minimise any downtime for remediation work done, so I thought I would ask on here for any suggestions on ways to alleviate the issue.

[ggrice]

I have a similar mixed environment with SRM and Veeam operating at the same time. I don't get the issue that you note although my SRM replication is not done via an SRA but rather via VMware replication so it does not use conventional VMware snapshots (but rather an under the hood SRM snapshot - apparently).

Another plus is that I'm using Tintri storage with the VAAI plug in so my Veeam snapshots create and delete with zero stun.

[saintdle]

Hi,

they are not using the SRA with the NetApp, just using vsphere replication. And I am sure this uses normal snapshots, nothing under the hood, as I've had to cleanup the stuck snapshots

I've upgraded to v8 today, so hopefully the snapshot hunter will help,

just wondering the best way to get around all this you see. Bit of a pain, as its not my company, I just turn up as a consultant for the customer.

[chrisdearden]

Vsphere replication uses a special type of snapshot - there is a little more info on it here http://forums.veeam.com/post64493.html

are they using VSS with their replication snapshots ?

you can't usually see the replication snapshots unless there is a prexisting snap on the machine.

what about using pre / post job scripts to pause and restart the replication ?

is the customer running Enterprise plus ? you might be able to get round a lot of the issue by using backup form storage snapshots which will minimise the impact of the vmware snap commits.

[chilli007]

Have you managed to figure out what's causing the issue between SRM and Veeam ?

[veremin]

Hi, Stan,

Do you mind answering several questions? What particular you get while working with combination of SRM and VB&R? Do you replicate or backup exact VMs via two solutions? Is one or two products taking care about VSS freez inside VMs?

More information regarding your situation would be helpful.

Thanks.

[chilli007]

Hello, thanks for the questions,

If VSS is enabled for SRM - even on just two of the 8 vms beings replicated - it appears to clash with the VSS that is enabled on VB&R jobs for all vms. Example, one vm not being replicated with SRM was constantly creating and removing snapshots of itself whilst replications - using vss - continued. As soon as they were paused the snapshot issue stopped.
We replicate 8 of the 25 vms we backup.
VSS was used on 2 of the replicated machines and VSS is used is used for the majority of vms in veeam.

[veremin]

I'd say if both solutions are trying to manage VSS freeze inside particular VM, conflicts are expected. If those VMs aren't running special VM applications, you might want to disable VSS processing in one solution or in the other one to avoid overlaps. Thanks.

[chilli007]

OK, thanks for the update, would you also say that we'd expect the same conflict within Hyper-V?

[veremin]

It is not related to underlying platform, but rather to the fact that if two independent solutions try to take care about VSS aware application present inside VM, some conflicts are expected.

[chilli007]

OK, thanks for your reply and support

[veremin]

You're welcome. Should other questions arise, don't hesitate to ask.

[saintdle]

Hi, im the OP here, no we havent managed to resolve the issue, or do further testing,

I was researching SRM, and it looks like you cannot set a blackout window for the SRM Replication jobs.

It may be that we re-architect the environment to allow for direct FC SAN backups, which should bring down the backup time