Backup the replicated VMs

[larry]

I have one Veeam server doing a offsite replication to my DR site. Can I have the Veeam server in the off site DR location backup the replicated VMs which has never been mounted ? I want both backups and a replicated VMs in the DR site without bring each VM over the WAN twice.

[Gostev]

Larry, yes you can do this, but you would have to disable changed block tracking in the job settings, as CBT is not capable of tracking external modifications to VMDK file. So using CBT in this scenario would result in bad backups. Also, note that because of disabled changed block tracking, the incremental backup speed will be a few times slower than it would have been with CBT.

[larry]

ok. thanks for the quick reply. Less than two minutes is pretty good.
For one VM it is over 500gbs, backups have gone from 6 hours to 20 minutes with change blocks. I will most like need to still pull that one over the wire twice. Any chance of replicating backups to a remote veeam server ? If I use VMcopy will that solve the change blocks? I am not sure of the differnt between copy and replicate anyway.

Totaly off the subject. Does Veeam care if I cut and paste from the user manual for a presentation to my IT and Risk committee's. Then I would like to use some of the manual for our procedure manual.

[larry]

I thought of an issue your post made me think about. If I went back in time with a SAN snap shot to a time before the last full backup or replication then does vspear store the change block information with the VM and will Veeam be ok with this. Should my policy be any SAN snap shot restore requires full backups that night ?

[Gostev]

Larry, for offsite DR our customers are typically using RSYNC or DFS-R to sync backups over to remote site. This works very well, and this is what I would personally do if I needed to keep my backups offsite. Based on the feedback so far, Linux RSYNC works best.

Here are some useful threads about this:
off-site backup ideas
Replicating .vbk Files Using DFS-R
Does anyone use RecoverPoint to replicate backups off-site?

Re: UG. We will never know you did that Seriously though, it's fine to do. But I find that some explanations are in UG are too technical and confusing for committee presentation. So instead, you could contact your Veeam sales rep, and ask to share with you some existing presentations.

Yes, CTK information is stored with VM files, so this will be fine. Veeam Backup will query latest changeID (USN), and because it will be lower than changeID memorized during latest backup, Veeam Backup will realize that storage rollback happened, and will not leverage CBT data for the given incremental pass - instead, it will scan the whole source image to determine blocks which actually differ from those stored in the latest state backup state.

[larry]

I use Rsync in our current setup. I been using Veeam to replicate to a remote esx and it works great. I also been using the Veeam server in the DR site to backup over our 100 meg WAN and that also works great. Both are much, much quicker when changed blocks are used than using Rsync and the Veeam reports are much better. Even using Rsync I would be pulling the data over the WAN twice. Unless I am missing something I see no reason to use any thing but Veeam for this.

This is what I am doing

I have two main sites, each are the DR of the other site. Both sites have production every day. My new DR plan I want to be able to mount the replicated DR VMs at will with little work. I am placing extra nics in all my production ESX servers at both sites. All ESX server will have their NIC setup the same with the same name. This way I can restore a VM in the DR site and the VM’s NICs will map to their normal NIC which is now on a DR network. The DR network would have no interaction with my production network. This will save me days of work to do a full test and I will be able to perform it during the day.

My main goal with Veeam was just to get a remote backup so I didn’t have the risk of shipping tapes. After using Veeam I keep finding new uses and DR will now be the most important.

[Gostev]

I have two main sites, each are the DR of the other site. Both sites have production every day. My new DR plan I want to be able to mount the replicated DR VMs at will with little work. I am placing extra nics in all my production ESX servers at both sites. All ESX server will have their NIC setup the same with the same name. This way I can restore a VM in the DR site and the VM’s NICs will map to their normal NIC which is now on a DR network. The DR network would have no interaction with my production network. This will save me days of work to do a full test and I will be able to perform it during the day.

I find this quite creative

The benefit and drawbacks of using replicas come from the fact that it is full-blown VM. As any normal VM, it does not need anything else to run - however, it takes 10 times more disk space and replication traffic (because Veeam Backup has to deal with uncompressed VMDK data). On the other hand, backup files are 10x smaller (so take 10x less disk space and replication traffic).

During your planning, consider that with version 5, you will be able to do the same thing (quoted above) with compressed backup files - without having to spend any time to actually restore VM from backup. Essentially, you will be able to just "mount" backup to ESX, and immediately power on some VMs stored in the backup file. So, whether you need this for VM recovery testing, or just ad-hoc access to data stored on those servers - this is ideal solution to my mind.

[egroeg]

[merged]

Hello all.

We have a DR site with LAN connectivity whereby we use Veeam to replicate approximately 50 vms daily to a standby ESX host.

We now would like to introduce a backup solution (to tape).

I am tempted to just request Veeam backs up my replicas' as opposed to running a fresh backup of the VMs at the primary site.

Can anyone argue why one wouldnt do this as opposed to conventional VM Veeam backup?

[Vitaliy S.]

The only drawback in this scenario would be an impossibility of using Changed Block Tracking because of the reasons explained above.

[jroesner]

[merged]

We have live vm's at an offsite data center, and are replicating them to an onsite server. I would like to take data offsite, and thinking about just backing up the replicatants to removable storage, looking for thoughts.

The theory being we would have a copy of our systems should all the servers fail.

(Or is there a better way to do this, no one wants to goto the data center every day to plug in media.)

[deesloop]

[merged]

I replicate some vms in our prduction environemnt to a DR site but want to
investigate backup side of things. I'm wondering if I could run a backup against
the vms in the DR area rather than against the production servers again? The
Veeam machine is in the DR site and I'm assuming it will be quicker to run the
backups against vms in the DR site rather than in the production evironment.
as the backups will be stored on disks in the DR site.

Also will I be able to restore files/emails from the backups of the DR machines
back into the production environment and/or the DR environment?

THanks a lot.

[pcrebe]

Sorry,
i don't remember if use or no CBT to backup a VM_replica that is naturally powered off.

Thank you,
Carlo

[tsightler]

I've tested this and it appears that a replica cannot use CBT, even if you enable it on the replica. Each time a new replica is pushed CBT fails on the next backup pass. Would be really nice if it worked.

[Gostev]

AFAIK CBT is generally not supported on powered off VMs (known VMware limitation). Although sometimes it works, generally it is not supported.

[daniflexx]

Hi All,

Now I'm deploying Veeam v6 on our vmware vCenter and I have a doubt,

I've read the uer guide, FAQ and the forum and I couldn't find nothing about that.

I want to create a replica job and a backup job of the same machines.

The question is: Can I back up the replica created in the replica job instead the production server, using CBT?

I want to reduce the impact of the backups to the patform.

Thank you.

[RH832]

Is there any problems with taking a Veeam replication and use Veeam Backup to backup the VM_replica? I am not sure if Veeam does anything with the replica that would cause a problem backing up by Veeam or vice versa.

[dellock6]

Never done it, but basically the Replica in v6 is a full VM, with some snapshots on it (those are the restore points).
I would say it could work, only be sure the replica job is not running while you backup that VM.

[RH832]

I created a job with CBT disabled as per other posts and it seems to work. Replication didn't complain either afterwards.

[Gostev]

Please note that with v6, you do not need to disable CBT any more, because the restore points are created and stored differently.

[trafsta]

Having read this thread fully, I am a bit confused what is recommended to be done under v6 (which we are currently evaluating). I am currently backing up to an offsite Veeam B&R DR server and have that backup job set to start a replication job to the same DR server (which is a Hyper-V server as well). This works well, but is there a better way than having to run the backup and then replication separately (which in turn hits the production source servers twice). What is recommended now with v6? For us the offsite Veeam server is across a gigabit link but we want to also in the near future try backing up and replicating across a 10mbit link in another city for extra protection, so I do not want to hit the production source servers 3 times if possible (still need to look into rsync as an option to possibly assist with all this...). Any insight would be appreciated.

[foggy]

If you need to have both remote backups and remote replicas of your VMs and do not want to put excessive load on your production, you could backup previously replicated VMs as described above in this topic. Just be aware of possible issues with CBT.

To replicate your backup files offsite for extra protection, you could consider using some file synchronization software (rsync is just one of them but seems to be the most commonly used one). Common solution among our customers.

[pcrebe]

Hi Gostev,

Please note that with v6, you do not need to disable CBT any more, because the restore points are created and stored differently.

Sorry but on Tue Jan 17, 2012 9:41 pm

AFAIK CBT is generally not supported on powered off VMs (known VMware limitation). Although sometimes it works, generally it is not supported.

So now for a v6 backup of a VM_replica may i enable or disable CBT? The VM is powered off but the job is set up for continue also in case of powered on VM_replica in case of HQ disaster.

It would be helpful if the best practice was added to the FAQ

Tanks,
Carlo

[trafsta]

Hmmm, yeah, I am still confused as to what to do in v6. I backup and then replicate to an offsite DR Veeam server (gigabit connection)... this hits the production servers twice (this is in a Hyper-V environment, not VMWare). Is there a better way to give me both a Hyper-V replica on the offsite DR Veeam server (which is also a Hyper-V server) AND backup files (which I'll be backing up to take twice a month with Backup Exec)? Keep in mind I only have the one server in the DR site so that is why I have Hyper-V, Veeam, and Backup Exec all on the one server... looking to get about 4 or 5 VMs replicated to it and in easily-bootable state in the case of a disaster, but also to have the nice backup files that I can send to tape once a month... currently I hit the production server twice to achieve this (after the backup job completes it triggers the replication job). I'm guess there isn't a better way to do this?

[Vitaliy S.]

Hi guys,

So now for a v6 backup of a VM_replica may i enable or disable CBT? The VM is powered off but the job is set up for continue also in case of powered on VM_replica in case of HQ disaster.

Based on our recent tests, looks like VMware has been able to fix that with their latest release, so CBT might actually work, however further research is still required.

I backup and then replicate to an offsite DR Veeam server (gigabit connection)... this hits the production servers twice (this is in a Hyper-V environment, not VMWare).

This possible CBT limitation does not relate to Hyper-V environments, since we use our own proprietary changed block tracking mechanism. With Hyper-V you can safely backup replicated VMs with CBT enabled.

Thanks!

[Gostev]

So now for a v6 backup of a VM_replica may i enable or disable CBT?

Yes, you should keep CBT enabled. If CBT does not work for any reason, it will not cause any issues (just slower backup).

[captainflannel]

So the recommended procedure for such a solution would be

Replicate Production VMs to DR Servers with CBT Enabled
Backup the Replicated VMs with CBT Enabled

Also incorporating windows to allow for VMs to Finish Backup and Disable replication during backup windows.
?

[captainflannel]

In VMWare environment, per my question. Not sure if Hyper-V would require different scenario.
Thanks.

[foggy]

So the recommended procedure for such a solution would be

Replicate Production VMs to DR Servers with CBT Enabled
Backup the Replicated VMs with CBT Enabled

Correct.

Also incorporating windows to allow for VMs to Finish Backup and Disable replication during backup windows.
?

If both jobs are configured on the same backup server, you can trigger the backup job as a post-job activity of a replication job, to perform that.

[captainflannel]

In a scenario where we are replicating to a second site, and then running backups of the replica VMs and then running a UAIR Exchange Restore, would we need to redirect the restore to the actual production server? or will UAIR just be looking for the FQDN of the original host/mailbox server? I have yet to try here wondering if anyone knew.
thanks.

[foggy]

Better to check, but I believe that if you specify the production server address in the Exchange AIR Wizard, it should be able to locate the corresponding replica VM in backups to perform the restore from.