Backup Clustered Virtual Servers in MS Hyper V

[pbarnard]

I am currently running in a Windows 2008 R2 64 bit environment utilizing Hyper V for virtualization. I am running TSM Server 6.2 and clients are also at 6.2.

In our stand alone environments, we have used the hyper v option to backup our virtual servers and this has worked fairly well.

However, we are in the process of clustering our virtual server hosts. What we have ended up with is a 3 server cluster and virtual servers running on any one of the hosts just depending on the situation.

So my question is, what would be the best method to backup our virtual servers?

Our first thought had been to install our backup client on each host server in the cluster and use the same method as before. However, we could end up with duplicate backups stored when a guest server moves from one host to another. This probably wouldn't be the best option.

Thanks for any help.

 

[moon-buddy]

I am guessing that by 'clustering' you meant that you are using a setup wherein cluster resources like File services is always available even if a host-controlling node fails.

This is no different than a normal cluster that uses physical rather than virtiual servers.

What you need to do is define a separate failover cluster resource for the shared disk or services which is then backed up by the active node.

 

[cmoeller]

pbarnard - Do you use individual volumes or a cluster shared volume?

In our case we do not use cluster shared volumes so maybe I can be of help. We don't use TSM's Hyper-V backup method either (instead we save the machines 'dirty' in filelevel) but I assume it will work the same or rather I cannot see a good reason why it should be different.

We have two opt-files, one for backing up the Hyper-V hosts' local drives and systemstates and one to save the clustered volumes. The cluster-opt has the "clusternode" and "clusterdisksonly" options set to yes.

What happes now is that when the backup is run with the cluster-opfile TSM queries which volumes are in fact clustered and backs them up - assuming the node name of the cluster. So no matter which node the backup is run on and where a virtual machine (and its volumes) "resides" at the moment there are no duplicate backups since it's backed up under the clustername.

I don't use TSM schedules either I have scheduled tasks on the machines. So I run a script that does a backup with the "local-optfile" and then with the "cluster-optfile" and I don't have to worry which resource is where.

Works flawless here, maybe you can give that a try.

 

[pbarnard]

Sorry I have not gotten back sooner, I was out of the area for a bit.

A little more detail as far as what we are doing is: We are taking 3 physical servers and utilizing clustered shared volumes and placing our virtual server vhd and vmc files on the shared volumes.

This is our first clustered server for any process, so everything is new to me.

I'm not certain what is meant by creating separate cluster failover resource. If this is referring to something along the lines of having one of the host servers backup the shared information natively itself, I was under the impression that since the virtual servers are active, they files will be locked from backup.

This is why were were trying to use the macro backup specifying the command "-vmbackuptype=hypervfull -vmlist=".

Of course I could just be entirely confused at this point.

 

[LOKI]

Hi,
we have some situation -Windows 2008 R2 Hyper-V in 4-cluster + CSV
TSM B/A client on each physical node
we use ASNODE param to collect data in one filespace
and it works... but...

If there is someone who know how to pass two backups at time
or better way how to backup please give me some clue.
Thanks.