Replica / DR architecture recommendation

[aaron@ARB]

Hi all,

I have setup my DR environment but I have some questions around what is the best approach in terms of utilising the bandwidth / lowering the transfer times.

I have two sites lets call them production and DR both with their own vCenters.

At the moment as I type this I have a replica being processed based off a seeded copy that is pretty much up to date (probably a few GB off but close) and its taking an awfully long time to run the replica copy. The replication job is running from the DR veeam console with the bottleneck being the source as reported by the job. I have a 100mbit connection between the sites and an IPSEC vpn that runs over that. Below is a picture of what I am seeing



Even though the bottleneck is source, given the processing rate is 7mb which is about what the ipsec VPN can throughput, is that really whats stopping it? Would I be better setting the jobs rather than pulling from the Prod -> DR site, would I be better to push the replicas from the Prod -> DR site instead? It just seems very very inefficient.

Any help / thought would be appreciated.

Cheers,
Aaron

[jdavidson_waters]

The job is busy “calculating digests”, it won’t do this on its next run.
It’s also not able to use CBT, it should use this on its next run.
= next run will be much faster. Be patient for this first run.

It also looks like you’re using the veeam Server as a single proxy. You should ensure you have a proxy in both production and DR sites and specificy this within the replication job.

[aaron@ARB]

Hi... yes I do have proxies (and veeam servers) on both sites, but it doesnt let me select the remote one, it just gives me the local ones when I select them. doesnt help that they're both technically called the same thing.

[jdavidson_waters]

have you added the production site proxy into your DR veeam install?
Can you share a screenshot of your DR veeam’s proxy servers? You should see the local/default VMware Backup Proxy as well as the production site proxy that you’ve added manually (can be added by hostname or IP).

[MichaelCade]

Hi Aaron, I presume you two sites "Production" and "DR" are both active sites with live workloads? This is why you have a Veeam management server at both sites?

The Veeam replication job will require a source proxy (closest to the production data) and a destination proxy (closest to the destination location)

if i am right in that you have an Active Active configuration then your destination management server should be the management for the replication i.e if "Production" holds your production workloads and you are wanting to replicate them to "DR" then the DR Veeam management server will control the replication and the production Veeam management server will only control the backup schedule.

Let me know if you have further questions, but hope that makes sense,

[skrause]

To add what the previous posters have said:

If you already have proxies/repositories in your Production site and your DR site, you can add them to your DR B&R configuration. Provided your load and scheduling aren't way out of whack, you shouldn't have problems having multiple Veeam B&R servers using the same proxies and repositories (I actually have three connected to mine).

You will want to have a repository at the source/production location to store the replica metadata per best practices (I use a small additional windows volume that is only mapped to the DR server managing replication).

With the link limitations you have, have you looked at using WAN accelerators?