Simultaneous Pull Replication and Backup

[ZackAngelo]

Is anyone using Veeam to do backup to disk locally and also pull replication to an offsite DR datacenter?

It seems that if a backup tries to take place during a replication interval it fails because it is unable to create a second snapshot.

[Vitaliy S.]

Hello Zack,

It is impossible to run both backup and replication jobs against the same VM at the same time, as all the VM files are being locked by the first process/job and you have to wait till the initial process is finished in order to use the files (VMDK, VMX, etc.) further.

Thank you!

[ZackAngelo]

But doesn't Veeam backup from a snapshot? Why does it need to lock all of the VM's files?

[Michael1337]

I had the same problem:

You can write a script to realize it !

[Vitaliy S.]

But doesn't Veeam backup from a snapshot?

Snapshot is taken to unlock parent virtual disk files, and as soon as this is done, the backup engine starts transferring source disks to your target storage.

[sjors]

Sorry to reactivate this thread, but I'm facing the same issues as is described here.

I'm using Veeam 5 to make daily local backups to disk.
(Those are then backed up to tape using ARCserve.)

We now have a backup site (new) connected through a 10Mbps link.
I want to replicate a number of VMs to that site.

The way I have set it up is by installing an appliance at the DR site on ESX 4.1 (single server).
(So this is a *new* server running Veeam; not the one doing my daily production backups.)

From that server, I do a pull replication from my main site.
The replication has intervals of 1 hour.
(This is done so that these jobs don't interfere with each other.)

The problem I have is that after the initial replication, the job kind of restarts with a full replication again.
As a result, each replication job takes like a day to complete, since everything needs to be copied over the slow WAN link again.
This also means that the jobs cross each other and everything fails.
(Yes, I have read that jobs cannot cross each other.)

My questions are:
1. Why does the job resync the whole VM instead of just changed blocks?
2. Am I missing something?
3. Is what I'm trying to do, even possible?

Thanks,

George

[Vitaliy S.]

Hello George,

Every subsequent replication job pass is incremental one. Could you please tell if you have any VRB files created on the target datastore? By the way what time did it take you to perform initial replication? Have you done it from the DR site?

Also assuming that you have full blown ESX host on the DR site, there is no need to perform pull replication, you can just specify service console credentials and run replication jobs from the source site.

Thanks.

[sjors]

Hi Vitaliy,

Yes, there is a VRB file in the target datastore.
The initial replication took about 20 hours, being a VM of 200GB.

At this point, the link is still 30Mbps, thus generating a transfer speed of about 3MBps in the replication job.
(This link will be reduced later to about 10Mbps.)

At the DR site I have a full ESX 4.1U1 host with one VM running Veeam.
The replication is pulled from this VM.

At the main site, I have a couple of ESX 4.0 and 4.1 hosts and one physical backup server running Veeam 5.
From this server I run daily (weekdays) backups of the VMs.

On the source ESXes I also see multiple snapshots (that are probably the result of mixed jobs.)

And I don't understand the last line, about the credentials?

George

[Vitaliy S.]

Yes, there is a VRB file in the target datastore.

Well...that means that you had an incremental run.

The initial replication took about 20 hours, being a VM of 200GB.

Is it an Exchange or SQL Server that you're trying to replicate? Do you have the same speed on all VMs being replicated?

And I don't understand the last line, about the credentials?

Here you go, existing thread that should explain you what service console means: Service Console Connections