Design Question

[jrp]

Hello,

We’re looking to replace DPM with Veeam v8 and are trying to better understand the architecture we need. Our setup is pretty simple:

Production Site:
All Hyper-V 2012 hosts
Dell R730 DAS only repository
Roughly 8TB raw vm data to backup

DR Site:
Hyper-V hosts
HP Proliant G7 DAS only repository
150Mbs line between sites

Goals:
1) Daily local backups (retention policy will be 7 restore points)
2) Daily backup copy jobs to store VM’s in DR site repository (retention policy will be 7 restore points)
3) Daily remote replica from backup jobs for handful of Tier 1 vm’s (retention policy will be 7 restore points)

Questions:
What’s the optimal placement for our VB&R server? In reading the user guide & various forum threads everyone seems to approach this a little differently. I see some people standing up their primary VB&R server in the local site to handle the backup jobs & then another VB&R server in the DR site to manage replication jobs, but my question is wouldn’t it be easier to just install a single VB&R server in the DR site to manage everything? In that way, if the production site goes down, you could easily restore from the backup copy jobs in addition to letting Veeam handle the replica failover process.

Is this an acceptable design, or is there something I’m missing? Is there a performance tradeoff with this single server in the DR site design? Would I be better off utilizing (2) VB&R servers instead?

Thanks in advance for the guidance. - jrp

[foggy]

What’s the optimal placement for our VB&R server? In reading the user guide & various forum threads everyone seems to approach this a little differently. I see some people standing up their primary VB&R server in the local site to handle the backup jobs & then another VB&R server in the DR site to manage replication jobs, but my question is wouldn’t it be easier to just install a single VB&R server in the DR site to manage everything? In that way, if the production site goes down, you could easily restore from the backup copy jobs in addition to letting Veeam handle the replica failover process.

People usually set up a local Veeam B&R for the purpose of fast operational restores in case something goes wrong (but not as seriously as the entire site loss). So having two Veeam B&R instances (one responsible for such restores in the local site and another responsible for replication jobs with automatic failover/failback in mind) is a typical approach.

[jrp]

Thanks for your response. If I may ask, why is a local B&R server faster in that scenario vs. a remote B&R server? In other words, if I had a single B&R server handling everything in my DR site, but was only trying to recover a VM in my production site, wouldn’t the actual recovery traffic only be between my local repository server & the local Hyper-V host (i.e. everything is still contained within my production LAN), or is there some other involvement with the B&R server that would still cause traffic to cross a WAN link & slow things down?

Again, thanks for the help, the Veeam approach is quite a bit different than what we’re used to with DPM so we’re just trying to ensure we architect things correctly the first time.

[veremin]

Windows FLR executed from remote console might be rather slow. So, it's recommended to have a cold console at the local site for the purpose of faster granular restores. Thanks.