Comprehensive data protection for all workloads
- Posts: 61
- Liked: 10 times
- Joined: Mar 01, 2010 5:57 pm
- Full Name: Glenn Santa Cruz
Looking for advice on setting up replication jobs, and I'm hoping you can share your experiences. I'm really debating aggregate jobs versus single jobs and trying to weigh the pros/cons of each. Firstly, we're talking about replication to a DR facility across a WAN with a relatively small pipe ( 10Mbps ); complete refresh from scratch (reseed) is painful due to the time involved -- we would like to avoid reseeds as much as possible.
For illustration, suppose we have the following recovery point objectives (RPO):
- 4 hours for 3 VMs: ( VM4-A, VM4-B, VM4-C )
2 hours for 3 VMs: ( VM2-A, VM2-B, VM2-C )
10 hours for 2 VMs: ( VM10-A, VM10-B )
I can see at least three strategies:
- Strategy 1 : define one job per RPO, placing appropriate VMs directly into the corresponding job. By "directly" I mean that each VM is selected within the replication job creation wizard. This would result in three replication jobs ( rep4, rep2, rep10 )
Strategy 2 : define one job per RPO, define a vCenter folder for each RPO, add the corresponding folder to each job, then move VMs into each folder. This has the advantage (presumably) of allowing for flexible RPO changes -- if I need VM4-A to suddenly have an RPO of 2 hours, I just move it to another vCenter folder.
Strategy 3 : define one job per VM
For both Strategies 1 and 2, does Veeam allow for movement of a VM into another job (either via direct selection or via folder manipulation) and *keep* the state of the replica? Or would we be required to reseed? Any pros/cons to the above? Maybe even other strategies that we've not considered?
- SVP, Product Management
- Posts: 29916
- Liked: 5852 times
- Joined: Jan 01, 2006 1:01 am
- Location: Baar, Switzerland
Glenn, moving VMs between jobs while keeping the state of the replica is not supported at this time - this would require reseed.
Given the specifics of your scenario (off-site replication via slow link), and the number of VMs to replicate, I would go with one job per VM, as this gives you more granular control.
Users browsing this forum: Google [Bot] and 23 guests