Backing up Hyper-V Physical host

[lmgtfy]

Please can anyone let me know what software they are using to backup their Hyper-V Physical hosts when they are part of a cluster.

 

We use Veeam to backup all VM's in the cluster which works well and have been using DPM to backup the hosts but since I applied some Windows updates during the Easter holidays to the hosts in the cluster DPM has been throwing up some VSS errors on the hosts and even causing the cluster service to stop on certain hosts failing the VM's to another host. I have had enough of DPM now and the way certain updates break it and need to change software before I go mad.

 

Can anyone suggest anything other than DPM for backing up the physical hosts that's works well with a cluster?

 

 

Cheers

[zag]

Maybe i'm missing something but why would you backup a physical host?

 

Surely its easier just to have a spare empty physical host ready to take any VM's restored using Veeam?

[lmgtfy]

Hi @zag your not really missing much, I guess its just for disaster recovery. We have 4 hosts and all the VM's could easily go on 3 without any issues at the moment. I guess it would save time though if the Host OS went corrupt for whatever reason and instead of rebuilding the server with all the configs and drivers, updates etc from scratch I could just blow an image and get backup and running?

 

So do you use Hyper-V or VMWare? Cheers

[zag]

Hi @zag your not really missing much, I guess its just for disaster recovery. We have 4 hosts and all the VM's could easily go on 3 without any issues at the moment. I guess it would save time though if the Host OS went corrupt for whatever reason and instead of rebuilding the server with all the configs and drivers, updates etc from scratch I could just blow an image and get backup and running?

 

So do you use Hyper-V or VMWare? Cheers

 

We use Hyper-V here.

 

Our disaster recovery procedure is designed to simply move VM's to spare host if we had a catastrophic failure of a physical machine.

 

I'd always recommend having a spare host set-up on-line to take the weight of any VM's, but you could also move them to another working host if there is enough capacity.

[fiza]

We use Hyper-V here.

 

Our disaster recovery procedure is designed to simply move VM's to spare host if we had a catastrophic failure of a physical machine.

 

I'd always recommend having a spare host set-up on-line to take the weight of any VM's, but you could also move them to another working host if there is enough capacity.

 

We use Hyper-V and do basically the same. We have 2 hosts running different VMs. Each backs up its VMs to an ISCSI connected QNAP. If one were to go down we would just bring up the VMs on the other host. Although both 'live' hosts neither is currently taxed too much with VMs.

[lmgtfy]

Surely some people backup their physical hosts though?

[plexer]

Surely some people backup their physical hosts though?

 

If all they do is Hyper-V what's the point?

 

I guess the only reason would be if you had a very complicated virtual switch setup.

 

Ben

[zag]

Surely some people backup their physical hosts though?

 

Nope, totally pointless imo

[tmcd35]

I agree, the time it take to re-install and reconfigure Hyper-V makes it pretty pointless. That said, if I was to do it I'd probably go down the route of a Ghost/Clonezilla/SCCM image.

[localzuk]

I don't bother backing up my Hyper-V physical hosts either. Even to build up a clustered environment it only takes about 20 minutes, so I can't see any advantage of doing a restore over simply doing a fresh Windows Server install and adding the Hyper-V role.

[lmgtfy]

@zag I'm not sure I agree. It depends on how complicated your setup is. I have a four node cluster and rebuilding a host would easily take me the best part of a day to get it to be identical to the other nodes in terms of software, updates, drivers, equallogic tools, veeam, ups software and config, iscsi connections. That's before I guess how to add it back to the cluster seeing as I have never had to do it.

[zag]

@zag I'm not sure I agree. It depends on how complicated your setup is. I have a four node cluster and rebuilding a host would easily take me the best part of a day to get it to be identical to the other nodes in terms of software, updates, drivers, equallogic tools, veeam, ups software and config, iscsi connections. That's before I guess how to add it back to the cluster seeing as I have never had to do it.

 

We always have a spare hypervisor running anyway (infact i'm paranoid so I have a couple), I'd just restore the virtual machines using veeam to an alternate physical host.

 

In an emergency situation like that, I think its worth the extra thousand quid to have spare capacity.

[lmgtfy]

@zag I wish I worked at your school I can only dream of a spare server but your right it would make it a lot easier :-) I should push management for a spare one even if it is of less spec CPU wise to bring it down to the thousand quid mark as last time we bought hosts they were around £3-4k each!

[Jobos]

So the general consensus is not backing up the host so to take one of my schools as an example, what would be the best method of backing up the VM's as I have a host that only runs hyper-v. How would I ensure the VM's are backed up without exporting or stopping the VM while the backup is taking place? There isn't a cluster just a sole hyper-v host.

[jamesb]

Can anyone suggest anything other than DPM for backing up the physical hosts that's works well with a cluster?

 

Your best bet is to have a base image ready to go - if you restore a backup then it's almost guaranteed that you'll get some confusion in the cluster. They tend to argue over VM ownership if you restore. If they fail you're better off forcing the machine out of the cluster, and building it back in from scratch as a new machine than trying to bring it back to life.

 

If you really do want a backup you may be able to get away with taking the host offline and running a backup with whatever you'd normally use, then bringing it back online. Whatever you do though try to avoid backing up any host while it's running VMs.

[localzuk]

@zag I'm not sure I agree. It depends on how complicated your setup is. I have a four node cluster and rebuilding a host would easily take me the best part of a day to get it to be identical to the other nodes in terms of software, updates, drivers, equallogic tools, veeam, ups software and config, iscsi connections. That's before I guess how to add it back to the cluster seeing as I have never had to do it.

 

Ah, that'd be slightly different for me then - I use pure Windows 2012 for clustering. No iSCSI, its all SMB 3.0. Setting up is super simple.

[localzuk]

So the general consensus is not backing up the host so to take one of my schools as an example, what would be the best method of backing up the VM's as I have a host that only runs hyper-v. How would I ensure the VM's are backed up without exporting or stopping the VM while the backup is taking place? There isn't a cluster just a sole hyper-v host.

 

BackupAssist or Veeam will do this perfectly. Doesn't require any stopping/exporting, backs up in the background using VSS.

[Jobos]

BackupAssist or Veeam will do this perfectly. Doesn't require any stopping/exporting, backs up in the background using VSS.

 

I'm using BackupAssist already with the method set to drive image aka bare metal. Is that incorrect?

[zag]

So the general consensus is not backing up the host so to take one of my schools as an example, what would be the best method of backing up the VM's as I have a host that only runs hyper-v. How would I ensure the VM's are backed up without exporting or stopping the VM while the backup is taking place? There isn't a cluster just a sole hyper-v host.

 

Veeam: Availability for the Modern Data Center

[win]

I don't bother to backup the host either. Make sure you're running 2012r2 on the host so you can export them without shutting the guests down.

 

Regular data backups with DPM, and then occasional full server exports to external USB drives.

[Davit2005]

20 VMWare hosts here in the cluster and we don't backup the hosts, all the hosts have local storage just for the ESXi in RAID 1 config, all Virtual Servers are on shared storage backed up by Veeam

 

And dare I say another +1 for Veeam.

[localzuk]

I'm using BackupAssist already with the method set to drive image aka bare metal. Is that incorrect?

 

For the VMs themselves? I have a "File Protection" job for that, however, bare metal would cover it too.

[Jobos]

I didn't know file protection would work while the VMs were running as in the past I shutdown the VMs and did an export but I could only do this in the holidays as it takes a while. The server is running 2012 but not the r2 version so exporting a running VM is not an option but it would be nice to have.

[win]

I recommend running the R2 upgrade. It will save you a lot of hassle in the longer run by exporting without shutting down first. Although it takes longer to export a live server so if you can shut down then do it that way.