Replica restore point slow removal / optimize setting of B+R

[emk23]

I have a ~2TB exchange 2007 server which I am replicating to our DR site.

The time which it takes to remove the 1 restore point is close to 1 hour.

My two data centers are connected at 1Gb, so I am also doing "backup" jobs to completely different storage/hardware at the DR site and taping that out (3month retention).

My backup and replication jobs are currently both configured to keep 7 restore points.

With that in mind, what should I do to optimize and reduce the time spent on restore points for replica jobs?

If I keep only 1 replica restore point would that solve it? Would that be recommended or not recommended given I also have "backups" on disk on separate hardware at the DR site PLUS taped out backups that go back 3 months.

[foggy]

Eric, when the oldest restore point is removed according to retention, this basically commits replica VM snapshot (consolidates it into the base disk), which obviously might take some time depending on the VM size (especially if the VM is hosting some highly transactional applications, like Exchange). You could pay attention to the target storage performance. Here are some tips regarding that.

If I keep only 1 replica restore point would that solve it? Would that be recommended or not recommended given I also have "backups" on disk on separate hardware at the DR site PLUS taped out backups that go back 3 months.

Being well-covered by backups, I think you can afford going with 1 restore point for replication.

[rschmaltz]

Hi -

We've got a replication job that runs relatively quickly until it reaches the phase where it removes the retention points. For whatever reason, applying the retention policy seems to take almost as long as the actual replication process. Is there something I can do to optimize or decrease the time necessary to apply the retention policy?

[Vitaliy S.]

Hi Robert,

What is your target datastore performance when snapshot removal operation takes place? What is your retention policy for the replication jobs?

Thanks!

[rschmaltz]

Hi Vitaliy -

I'm not sure how to answer your question in regards to performance. The datastore is one of NetApp's new Virtual Filers running on a dedicated IBM x3650 M4 with DAS 10k SAS drives. Other than replication jobs, this filer doesn't have a lot of activity. We do a nightly replication which keeps the last 5 copies.

[Vitaliy S.]

You can review performance of the datastore (IOPs, latency) via free version of Veeam ONE or check it in the vSphere Client.

[bhagen]

I have several replication jobs running once a day. All jobs are setup for 7 restore points.

One of the jobs - my exchange server replication job - consistently takes between 1hr and 1hr30m to finally say "1 restore point removed by retention policy".

The other 6 jobs take between 1 and 10 minutes to remove the restore points for each VM. (One of these jobs has a 2016 server with a 15TB drive, and it takes about 2 minutes to remove the restore point.)

Is this a known issue for exchange servers? The exchange VMs are Windows Server 2012 with Exchange Server 2013, and they each have 4 drives; 200GB, 200GB, 700GB, 4TB.

If it's not a known issue, where should I be looking to troubleshoot this?

[foggy]

Might be expected considering the VM size and such a highly transactional app as Exchange. You can find some hints in the thread above.