Backup Copy Job over IPSEC - throttled, killing Exchange

[Tumac]

We are running a Backup Copy job of our VMWare environments between 2 sites.

150Mbps connections at both sites
Veeam WAN accelerators in use at each site
During business hours, Veeam network throttling enabled and capped at 2MB

Exchange hosted at corporate for Branch site. Most users are on RDP desktop terminal services so cannot run Outlook in cached mode.

When our backup copy is running, Exchange takes a beating and users connections freeze and slow down. There is some ebb and flow, at times it is unusable and other times it is bearable. Cached mode Exchange users are fine. Remote access to servers RDP is fine (a little slow but not unbearable). Internet browsing is fine.

We thought we had an issue with the firewalls (still may). Firewalls show IPSEC bandwidth utilized at 2MB. Throughput on our NICs will fluctuate between 4 and 20Mbps. Rebooting firewall would alleviate for a while.

Veeam backup copy jobs also run slow (or perceived this way) - we are pushing on average 4-5 GB branch to corp and 30gb from Corp to Branch. Replication takes 8-24 hours depending. Not sure of actual transport time vs merge time - running Reverse Incrementals

I have a hard time believing we are doing anything out of the norm of a larger enterprise.

Help - what can I clarify?

[veremin]

Could you elaborate on the issue further? Backup copy job has nothing to do with production VMs working with backup data only, so, hardly can it be the root cause of Exchange server freeze, ping loss, etc.

Or the problem is that while running a backup copy job floods the network, not taking throttling rules into account?

Thanks.

[dellock6]

From the informations posted, the Exchange problems can be because exactly what you said: you are capping the bandwidth, Veeam ir respecting the capping, and because of this as you said "Replication takes 8-24 hours". For this entire time, the VM snapshot is open and grows probably in size. So at the end of the replica, when the commit happens, this crates the well known issues to the Exchange VM.

I'd prefer first to confirm this is the case, before moving to give you some tips.

Luca