Preupgrade procedures discussion

[k00laid]

Hello all, I am preparing to try and roll my production environment to v6 late this week and wanted to get some input on design prior to actually doing it.

My existing setup is 4 ESX servers with vCenter at HQ backing up nightly through v5 to a Dell PowerVault NX3000 running Windows in Reverse Incremental mode. These jobs run locally starting at 6 PM and have at least 1 GB of bandwidth available to them so the jobs tend to finish up at 8 PM.

I then have an identical Dell PV NX 3000 at our DR site that runs the exact same jobs nightly, just in traditional Incremental mode with synthetic fulls on Saturdays. These start at 9 and when things are going good finish at 2 or 3 in the morning. There is a 50 Mbps pipe between HQ and the DR sites exclusively for this purpose. At best this setup is mostly stable, anytime we add a VM or extend a Virtual Disk jobs have to be juggled or they begin to fail spectacularly.

What I am hoping for with v6 is to upgrade is to basically do all my backups locally and then duplicate them using a Veeam controlled process to the DR site. Both servers would be running v6 and as the jobs are duplicated they would be automatically imported into Veeam at the DR site. After the file copy process each site would be able to process its own synthetic full process keeping us from copying the full file every weekend.

I also have high hopes that each individual job will now have the capability of using the whole pipe. In v5 if two jobs were running concurrently I could reach the 40 Mbps level to our DR site, but if only one job was running it was only able to hit 20 Mbps.

Is what I describe above doable or am I dreaming? Any particular recommendation for making this process to work as efficient as possible?

Thanks,
Jim

[k00laid]

I should also add that the DR site job is currently running through a CentOS VM with the NX 3000 mounted in order to make the synthetic full work correctly.

[Vitaliy S.]

What I am hoping for with v6 is to upgrade is to basically do all my backups locally and then duplicate them using a Veeam controlled process to the DR site. Both servers would be running v6 and as the jobs are duplicated they would be automatically imported into Veeam at the DR site. After the file copy process each site would be able to process its own synthetic full process keeping us from copying the full file every weekend.

Assuming that I have fully understood your scenario, the biggest drawback here would be that you cannot map backup files between backup servers at this time. You can only map backup if it was produced by another job on the same server. That said, I would still recommend you to use single backup server which will be backing up to two different locations, but do not forget to place target Windows repository on the remote site to keep the traffic of synthetic backup rebuild locally.

[k00laid]

As I understand it I don't need to run this through a Linux host any more to make the synthetic fulls work correctly? I can backup directly to the repository?

How would I handle getting the jobs into the offsite Veeam server (installed on the same box as the offsite repository) for testing?

[Vitaliy S.]

As I understand it I don't need to run this through a Linux host any more to make the synthetic fulls work correctly? I can backup directly to the repository?

Yes, that's correct. All our repositories will have tiny agents deployed for these procedures.

How would I handle getting the jobs into the offsite Veeam server (installed on the same box as the offsite repository) for testing

Jobs cannot be imported, but you can import backup files in the same way you did with v5.

[habibalby]

Replicate on daily and initiate your backup from DR site. That's what I'm doing.

Thanks,

[k00laid]

That's what we do now but the performance is horrendous.

[habibalby]

Do you select WAN Method for the VM?