Local Hyper-V backups with 15-minute RPO, should we do CDP?

[dimaslan]

We are working on a proposal for a prospect customer with about 50 VMs which will be moved from VMware to Hyper-V and we're designing a resilient Veeam solution for them
They have about 12 VMs (will be on Hyper-V by that time) which are being backed up with another solution every 15 minutes now.
I have seen vmware taking a very long time to create a snapshot and veeam to assign all the components and I feel the 15-minute RPO will marginally be met.
Should we consider CDP? Is CDP a thing only for remove replication? I am seeing CDP was not available for Hyper-V. Should we consider Agent for Windows instead of relying on Hyper-V snapshots?

[mjr.epicfail]

Hi,

CDP will not work here as you already found out that CDP is a VMware only feature.
I would also stay away from the agents unless you really have to.
Its better to understand why it would take so long to take a snapshot and fix that if possible.

[david.domask]

Hi Dimitris,

Would second Danny's comment, as CDP is only available for VMware at this time.

What roles do these servers have? That is, the low RPO is because they host applications like SQL (any) or similar roles?

The reason I ask is that if it's SQL data, daily backups of the server itself combined with Transaction Log Processing can help by backing up the relevant application data separately from the image level backup, meaning you don't need tons of snapshots each day to protect the SQL data. Similarly, you can use Veeam Plug-ins for Enterprise Applications to similarly avoid tons of snapshots on the production server. The data is still protected and you have full recovery, but the important data will be protected without constant snapshotting.

Since you mention the VMware processing took a bit longer than usual and the plan is to migrate to HyperV, have any tests been done with how HyperV performs for these backups? Long snapshot creation time and resource assignment would need to be investigated by Veeam Support, but if I get it right this issue was only identified for the machines on VMware, am I correct? I think maybe doing a test restore without connecting the restored VM to any networks then testing the backup performance on HyperV would be ideal just to establish a baseline. There's nothing inherently wrong with backups every 15 minutes, but it can be quite resource intensive due to the snapshots (checkpoints in HyperV), and it might make sense to protect the data via other means if feasible to reduce the load on the production environment.

But since there will be a migration anyways, it would be great to test first and see if it's really an issue you'll need to face after the migration, as the backup process differs slightly between Vmware and HyperV.