Host-based backup of VMware vSphere VMs.
Post Reply
cerberus
Expert
Posts: 171
Liked: 19 times
Joined: Aug 28, 2015 2:45 pm
Full Name: Mirza
Contact:

Switching replication job source from "backup files" to "production storage"

Post by cerberus »

Hello,

Our backup jobs run every 24 hours, and the replication job is chained off that (using from backup files as the source repository) that replicates the VMs to a DR site within a few hours, and I am looking for ways to lower the RPO at the DR site.

Looking to switch the replication job to continuous schedule, where the source repository it uses is from production storage (actual VM state), with "backup from storage snapshots" integration enabled, underlying storage is Pure FA, and vSphere CBT enabled.

Are there any live VM performance concerns when using production storage as source? If we are using storage snapshots, should I be concerned with "snapshot stuns", or is that not at play with storage snapshots?

Since we are no longer replicating from backup source, where the backup job would have had the "VMware Tools quiescence" and "application-aware processing" features enabled, I see these as an option now on the replication side, given that this replication will run every few hours, I am a bit concerned that this could also introduce some "blips" to the live VM.

Anyone out there with continuous replication jobs, where they are using production live VMs as source?
david.domask
Veeam Software
Posts: 2863
Liked: 657 times
Joined: Jun 28, 2016 12:12 pm
Contact:

Re: Switching replication job source from "backup files" to "production storage"

Post by david.domask »

Hi cerebus,

There will be a brief VMware snapshot as described in the Storage Integration User Guide; ignore that it tells about Backup here, the same applies for Replica from storage snapshot as well.

So for applications that are exceptionally sensitive to snapshot, you may still see a blip, but for most workloads it should be fine.

One note on switching the source from Backups to Production; this will trigger "digests calculation", a full scan of the replica for the first run after switching; digests calculation may take a bit of time and regrettably not many ways to speed it up. Creating the replicas fresh will avoid this but naturally will require a full pass, but since you're going to be using replica from storage snapshots, the impact on production should be quite minimal if any.
David Domask | Product Management: Principal Analyst
Post Reply

Who is online

Users browsing this forum: No registered users and 4 guests