Concurrent Veeam Replications

[bproteau]

We will be using Veeam as part of a Datacenter to Datacenter migration strategy. We have 10GbE link (~30ms latency). Source is made up of 20 ESXi 4.1 U1 Hosts sharing about 30 datastores. Destination will be similar. I'm planning on spacing VM jobs (1 VM per job) on serveral Veeam servers where when 1 job finishes, it will kick off the next job etc.

Since this is Snapshot based, I am wanting to figure out what concurrent job limitations are. I assume these are all VMware based so:

(1) What is the concurrent snapshot recommendation.
(2) Is this limitation, per Host, per datastore, or per datacenter?

[Gostev]

Is your production storage VAAI-enabled?

[bproteau]

Destination will be VAAI enabbled but, source is not.

Destination is 3PAR source is EVA.

[Gostev]

For non-VAAI enabled storage, the recommendation is to keep the number of snapshot to a minimum to reduce perfomance impact from SCSI reservations.

For VAAI-enabled storage, SCSI reservations are offloaded to storage (and performed on block level vs. LUN level), so there is no limitations.

[bproteau]

I figured it was a minimum. I want to translate that into an actual number for my source environment. How would I calculate it?

[Gostev]

Unfortunately VMware does not specifically define "minimum", otherwise I would have included the number. All they say is "keep it to the minimum during production hours to reduce performance impact of SCSI reservations".