Backup Copy Job WAN Accelerator v7

[beovax]

Looking for some advice on the recommended/supported way of configuring a backup copy job for off-site using the WAN Accelerator.

The Backup jobs are split into various jobs and run at different times. Reading the configuration guide on the WAN accelerator it mentions that you cant use the same source WAN accelerator for multiple copy jobs. So if I want to mirror the repository from primary site to DR site how do I set this up? As the recommended method with the WAN Accelerator seems to suggest a single backup copy job with all the VM's in, which creates a single file in the DR site with all the VM's within which is rather large. If this file gets corrupted or there any issues would lose all the VM's contained. Ideally wanted to have separate backup copy jobs to mirror the repository and utilise the WAN accelerator.

Any recommendations on how this should be configured?

[foggy]

Reading the configuration guide on the WAN accelerator it mentions that you cant use the same source WAN accelerator for multiple copy jobs.

Could you please clarify regarding this, as actually the same pair of WAN accelerators can be used in any amount of jobs. The only limitation here is that the source WAN accelerator can process a single task at a time only.

So you can create any required number of backup copy jobs (either connecting them to the backup jobs or backup repositories) and specify either the same pair of WAN accelerators or utilize additional servers as WAN accelerators if needed (but keep in mind space requirements for the global cache).

[beovax]

I read the step 10 in the help file under the section Specify WAN Optimization Settings.

IMPORTANT
You cannot assign one source WAN accelerator to several backup copy jobs that you plan to run simultaneously. The source WAN accelerator requires a lot of CPU and RAM resources and cannot be shared by a number of backup copy jobs. As an alternative, you can create one backup copy job for all VMs you plan to process via one source WAN accelerator. The target WAN accelerator, however, can be assigned to several backup copy jobs.

[veremin]

The key word there is “simultaneously”. May be the exact wording sounds a bit confusing, but what implies there is that source WAN accelerator will only process one task at a time, while other tasks will queue. Thanks.

[beovax]

OK thanks for clarification. As the backup copy jobs are all scheduled for the default times etc I assumed these would be running simultaneously rather than in a queued fashion.

[veremin]

Assuming that they use the same source WAN accelerator, they will be executed simultaneously, but WAN accelerator will process only one task at a time. Thanks.

[LinkeLoe]

I came across this post when looking for some best practices in our situation. We have the same kind of setup: multiple backup jobs, running at different times, targeted at different repositories. I'm trying to get a backup copy job running in such a way that it runs in a proper time window. However, we have backupfiles of 500 to 900 GB in size. What would be a best practice to setup backup copy jobs? We have a 50 Mb/s connection between the two site, so I guess I have to seed the backupfiles to the offsite repository. Should I create the same multiple repositories at the DR site, or should I create just one? Should I create multiple backup copy jobs, or just one?

[foggy]

There's no need to have multiple repositories on target and, typically, it is enough to have a single backup copy job for all your backup jobs. Are you going to copy backups offsite on a daily basis?

[LinkeLoe]

We are backing up our VM's on a daily basis and we would also like to copy these backups daily, but I'm not sure we can do this. Most of our servers are between 50 and 200 GB in size, copying backups of these servers is no problem at all, but we also have some servers like Exchange and SQL servers, that are between 800 GB and 1 TB in size. The initial sync definitely won't make it in time, so I guess I have to seed the target repository.

I am now making a backup copy locally on the same server as the Veeam backup server. If this copy's finished, perhaps I can copy the backup file to USB disk (if I have one that's large enough) and ship it to our remote site, but I'm afraid this whole process will take at least two days and the backup copy will be too old...

The local copy job now gave me a .VIB file of 900 GB, but the job skipped two VM's because there were no up to date restore points available. I am now running a backup job which includes these VM's and will let the copy job re-sync after this. I guess that, after a full synchronization Veeam will perform transformation on the .VIB file, so that I have a .VBK file which includes all VM's, that I can ship to the remote site. Is this correct?

Can you give me some best practices on how to set this up easily?

[veremin]

I guess that, after a full synchronization Veeam will perform transformation on the .VIB file, so that I have a .VBK file which includes all VM's, that I can ship to the remote site. Is this correct?

Minimal retention policy that can be set to backup copy job is 2. So, assuming that one restore point (.vbk) already exists, the next restore point will be an incremental one (.vib), and no transformation activity will be performed, untill the chain will reach, at least, 3 points.

Probably, the best approach will be to wait till the scheduled full backup takes place, then, copy resulting file to remote location via USB device, and use it later as a seed for backup copy job. The description of the seeding procedure can be found here.

Thanks.

[LinkeLoe]

Probably, the best approach will be to wait till the scheduled full backup takes place, then, copy resulting file to remote location via USB device, and use it later as a seed for backup copy job. The description of the seeding procedure can be found here.

I know how to seed a file for the backup copy job. The problem with this is, like I wrote before, that we have several backup jobs running, targeted at several backup repositories. This results in several backup files, that need to be combined into one file during backup copy.

[veremin]

Then, creating a local backup copy job and using resulting file as a seed (as you have already done) seems to be the best option. This way, only the difference will be transferred to target location. With the existing connection (50 MB/s), the backup copy job will be able to copy the difference during a 24 hour backup window.

Thanks.

[LinkeLoe]

I think I will let it run locally first, for a few days to see what happens. The total size of the initial backup copy is 1.6 TB, last night Veeam produced a 181 GB VIB-file. It is now working on transformation (I already had a backup copy, which had only 2 small VM's inside), but It's doing that for 3 hours now and it only got up to 41% yet...

[veremin]

Yep, you will have to wait until transformation is finished and full backup file is moved forward in the backup chain. The resulting .vbk file can be later used as a seed for backup copy job. As mentioned in the referenced thread, backup copy job can’t be mapped to the chain that already contains increments – only to a single full backup.

Thanks.

[LinkeLoe]

Right... The transformation's finished. The result is a .vbk file which is 2.5 TB in size and a 190 GB .vib file. Now that I know this, I'm gonna think of a way to ship this file to a server off-site...