Backup the replicated VMs

[captainflannel]

Hopefully not to off topic, regarding to the licensing for such a solution of Replicaiton and then backup of Replica vms. Would we need socket licenses of the source ESXi and Replica ESXi? Not sure if we need to have the ESXi servers to have licenses which are sourced for replica as well as source for backup. This would potentially double our licenses need if so.
Thanks!

[Gostev]

Yes, good point. You do need to license every host where VMs you are backing up are registered. Thanks!

[mdornfeld]

[merged]

Anton already pointed out that it can be done in his FAQ:

Q: Can I backup or replicate a replica VM?
A: Yes.

I'm interested from the standpoint of not hitting my production disk as hard if I'm backing up off of the replica. I was just curious about a few things:
1. Is this a fairly common practice?
2. I assume this could work, but wanted to verify. Could a backup job being running off a replica at the same time a replica job is updating the replica?
3. This is more of a VMware functionality question, but if I restore a backup job from a replica that has multiple snapshots on it, do I get all of the snapshots, just the oldest (base) snapshot, or just the lastest snapshot?

Thank you!
Matt

[Gostev]

Hi Matt,

1. Yes
2. Yes
3. Just the latest state

Thanks!

[dprescutti]

Vitaliy:

So has there been any further testing done on cbt backup of replicas?

You mentioned "lastest" release by vmware - do you mean esxi 5.0?

I'm definitely running into this issue with CBT failing v6.1 backups of replicas in esxi 4.1. The ctk.vmdk files aren't even created in the replica folder.

Thanks,
Dave

Hi guys,
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.
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!

[Vitaliy S.]

Dave,

Yes, we have done a further research on this topic and it appeared that VMware CBT cannot be used for replicated VMs. As you've correctly stated, VMware does not create ctk files for powered off VMs, that is why our jobs automatically failover to our proprietary change tracking engine, so that only changed blocks are retrieved by your backup job.

Thanks!

[dprescutti]

Thanks for the quick response. Has there been discussion about enhancing the proprietary change tracking engine to speed things up?

[Vitaliy S.]

Well...we have to scan each and every block and then verify this block presence in the backup file, which obviously takes time. Also since replicated VMs are powered off and do not produce any load on your production storage, running backup jobs a bit longer then usual should't be a big issue for you, right?

[dprescutti]

It's more about fitting it into a backup window for full offload to tape daily at the DR site. If there are no plans to improve this I'll plan accordingly.

Thanks again!
Dave

[Vitaliy S.]

Ok, thanks for the additional clarification.

[sve]

Hi,
Does anyone has tried or experience with taking backup of a replicated VM (via veeam replication job) ? (veeam 6.1 version)
My test(s) gave me failure responses - my guess is that the snapshot used for replication deny the backup of mounting the disks ?
Is my assumption right ? Or can this be solved ?

Thx

[foggy]

Stefan, backing up of replicated VMs is possible, please review this thread for details and limitations. If you experience issues while doing that, please contact our technical support team directly. Thanks.

[odge]

While backing up a replicated VM works, is there no way to disable the warnings?

I know CBT on this job wont work, I dont need to see the warning for it etc.

REgards

[veremin]

While backing up a replicated VM works, is there no way to disable the warnings?

I know CBT on this job wont work, I dont need to see the warning for it etc.

REgards

I must be missing something, but why not just disable CBT functionality as a whole in this particular case, as we know for sure that CBT doesn't work while backing up VM with at least one snapshot? Thus, there will be no warnings related to CBT not working properly and so forth.

Thanks.

[odge]

I'm looking for an option to do that outer job, not sure how...

[tsightler]

In the job settings, on the storage options, click Advanced...vSphere. The options for disabling CBT are on this screen.

[odge]

Foolishly I've followed the wrong thread.

I've been reading a backup thread, but actually I'm using a copy job. I'll start/find another thread for the copy job

[veremin]

Foolishly I've followed the wrong thread.

I've been reading a backup thread, but actually I'm using a copy job. I'll start/find another thread for the copy job

VM Copy Job doesn’t use CBT at all. However, you can try to put into use the following Power-Shell script and see what will happen in this case.

Note: If after implementing this script cbt warnings disappear, it’ll mean that VM Copy Job is trying to use CBT despite of the fact that such behaviour isn’t expected.

Code: Select all

asnp VeeamPSSnapin
$Job = Get-VbrJob -name "Name of your Job"
$JobOptions = $Job.GetOptions()
$JobOptions.ViSourceOptions.EnableChangeTracking = $False
$JobOptions.ViSourceOptions.UseChangeTracking = $False
$Job.SetOptions($JobOptions) 

Hope this helps.
Thanks.

[odge]

Good idea to try, will give that a go, thanks (next run of job on Sunday)

[Yuki]

Hi all,
I've tried to search forum, but can't seem to find the right topics. We are still evaluating if it would make sense for us to attempt making backups from replicas over the WAN so we don't ship the same data twice and instead make our replication more frequent.

Can someone please provider information on potential issue and pitfalls of doing this? Links to other threads on this topic are appreciated.

[Yuki]

Part 1:
Ok, i'm moving one of our big servers to backup from replica at remote site. This gives us more time to run backups and we only copy data off-site once. However, i'm curious about the earlier statement that we can be updating replica at the same time the backup of that replica is running. How exactly does this work?

Scenario 1:
Replica kicks off first - VMDK files are being updated live on target server (?)
Backup starts - is it going to backup inconsistent VMDK that has partial data?

Scenario 2:
Backup kicks off and starts backing up STATIC vmdk
Replica starts and begins updating VMDK as backup process is reading it

How is it that they do not interfere with each other?

Also - our file server has 5x2TB VMDK files that are spanned inside guest OS into a single 10TB volume (dynamic disk). Could we run into situation where one VMDK is being updated while backup captures another VMDK that has yet to be updated and the end result in the back that one VMDK was changed by replica and one wasn't? I guess the same question is valid for a reverse scenario.

I see a lot of benefit in backing up cold replicas and would like to understand the process in better detail before committing to it. Really would like to prevent a situation where we set it up and in case of emergency find out that there was an oversight on our part in how exactly this works and what the limitations are.


Part 2
We are already running replica from production to on-site secondary server and from production to off-site tertiary system. I would like to run replica from secondary (on-site) to tertiary (off-site) and increase the frequency of the on-site replication.
System1 - production (on-site)
System2 - secondary replica (on-site)
System3 - tertiary replica(off-site)

Currently, System2 and System3 hold "server_replica". Any way i can map system2 replica to system3 replica without it making a new "server_replica_replica" ? I haven't tried this, but assume thats what will happen if i just setup a new clean replication job.

[Vitaliy S.]

Hi Yuki,

Scenario 1:
Replica kicks off first - VMDK files are being updated live on target server (?)
Backup starts - is it going to backup inconsistent VMDK that has partial data?

Scenario 2:
Backup kicks off and starts backing up STATIC vmdk
Replica starts and begins updating VMDK as backup process is reading it

How is it that they do not interfere with each other?

There will be no inconsistency, since snapshot will be taken for all disks at the same time.

Could we run into situation where one VMDK is being updated while backup captures another VMDK that has yet to be updated and the end result in the back that one VMDK was changed by replica and one wasn't?

No, you can't run into this. It's the same as above... snapshot is taken for the entire VM with all disks. It doesn't matter what happens to the VM disks during backup job run, as you will be backing up VM data from the snapshot that was taken by the backup job/replication job.

Part 2
We are already running replica from production to on-site secondary server and from production to off-site tertiary system. I would like to run replica from secondary (on-site) to tertiary (off-site) and increase the frequency of the on-site replication.
System1 - production (on-site)
System2 - secondary replica (on-site)
System3 - tertiary replica(off-site)

Currently, System2 and System3 hold "server_replica". Any way i can map system2 replica to system3 replica without it making a new "server_replica_replica" ? I haven't tried this, but assume thats what will happen if i just setup a new clean replication job.

Yes, it should work.

Thanks!

[Yuki]

Another question:
We are running backups from VM replicas at DR site. During the job we get a message saying that CBT can't be used due to snapshots on the replica (we maintain two weeks worth of snapshots). If we reconfigure to only maintain 1 replica, so no snapshots are present, then we should be able to use CBT. But from what i read in this thread - it will be Veeam's own CBT and not vmware's ctk file based.

Keep the above info in mind, i have 3 questions:

1) If we keep 1 snapshot - we can use Veeam's CBT, right?
2) will we see any performance improvement over not using any CBT engine at all?
3) there is no point in configuring application aware image processing or indexing since machines are powered off, correct?

[veremin]

Whenever VMware CBT cannot be utilized, it is Veeam’s proprietary filtering mechanism that is being used.

In general, VB&R scans through the VM image and a checksum (hash) is calculated for every data block, then it verifies what blocks have been changed on your VM compared to the VM image stored in the backup file.

This mechanism doesn’t care about snapshot presence at all. However, full scan of source VM is needed to determine blocks changed since last pass, basically, it’d be the same amount of time required to perform Full Backup, though, in this case VB&R would transfer only changed block.

Yes, Application Aware Image processing setting, as well as, indexing one could be unticked in case of powered off VM.

Hope this helps.
Thanks.

[Vitaliy S.]

We are running backups from VM replicas at DR site. During the job we get a message saying that CBT can't be used due to snapshots on the replica (we maintain two weeks worth of snapshots). If we reconfigure to only maintain 1 replica, so no snapshots are present, then we should be able to use CBT.

Not exactly, even with 1 restore point specified, you will still have a snapshot present on the replicated VM.

[kamalkant]

I have replicated my virtual machine with Veeam backup and replication 6.1.
problem scenerion 1..My VM had 4 different virtual disk ( total size was 270 gb)
Now what has happened, I have started my replicated VM, that has only one virtual disk size (270 GB)..
how could I have 4 separate disks is replicated VM?
..........................
Problem scenario 2.
I am running a backjob for my replicated Powered on machine.. and Let me tell you the the progresss:
1) first full backup ran successfully.
2) 2nd incremental backup jon ended up with error:
"Cannot use CBT: Soap fault. A specified parameter was not correct. . deviceKeyDetail: '<InvalidArgumentFault xmlns="urn:internalvim25" xsi:type="InvalidArgument"><invalidProperty>deviceKey</invalidProperty></InvalidArgumentFault>', endpoint: ''
Unable to truncate transaction logs. Error: KeepSnapshot error: [Freeze job already stopped.]"
3) retry failed with below error:
"Error: Oib already exists in point 'id=89a1e023-b868-408a-870d-82d784f513e5:type=Basic:alg=Syntethic:creation_time=3/15/2013 11:31:33 PM'"

VM has two active snapshots now. why are they not being taken care by the commands sent by VBR 6.1.
My vmware version is 4.1... Please provide some resolution

[Vitaliy S.]

how could I have 4 separate disks is replicated VM?

Not sure I understand the question, can you please clarify it further? Do you see all disks replicated even though you've excluded them from replication job?

As to the second issue, then be aware that with all technical issues you need to open a support case with our technical team as it described when you click New Topic.

Thanks!

[kamalkant]

Hi Vitaliy... befor replication my VM had 4 virtual disk connected to it , but after replication only one virtual disk comprising the total size. there is no issue with the volumes seen on windows server. But my point is why replicated VM segregated all 4 virtual disk to one virtual disk? why not 4 different virtual disk as in the source VM?

[Vitaliy S.]

But my point is why replicated VM segregated all 4 virtual disk to one virtual disk? why not 4 different virtual disk as in the source VM?

Veeam B&R does not have this functionality and simply cannot merge virtual disks into one. Can you please tell me how do you check this?

[kamalkant]

Hi Vitality,

you are right. I am sorry , I got confused between two different VM. No issue with virtual disk.
for other issue related to CBT , I am going to open a case shortly. Thanks again.