Planning and consideration for replicating Production VM ?

[albertwt]

People,

I've got Veeam Backup & Replication softdware v 8.0 patch 2 (Enterprise Edition only with 22 CPU sockets license) and also VMware vSphere 5.1 U2 (under single Data Center) with about 200 VMs running in my production environment.

The physical Veeam backup server is running on the head office (performing Disk to Disk to tape) and I'd like to plan for a secondary DR site in another physical Data Center site, my questions are:

1. Do I have to double up my Veeam license to match the CPU socket like-for-like (eg. 2x22 = 44 CPU sockets) for the standby site ?
2. Do I need to have Veeam Enterprise Plus with WAN accelleration feature to do vSphere VM replication for DR site ? or can I still utilize the existing hardware appliance from 3rd party like Riverbed ?

Any kind of help and suggestion would be greatly appreciated.

[chrisdearden]

As long as you are only going to replicate to that standby site , you wont need any additional licencing for it. You only need to licence the source for a backup or replication job.

you might not need WAN acceleration if you have sufficient bandwidth between your sites to replicate the workloads with a frequency that matches your required RPO.

If you have general purpose WAN accelerators in place like a riverbed you may find that unless they are specifically for Veeam backups , that they cache just fills up with Veeam data , displacing the data you actually want cached for general operations.

If you look at the size of your incremental's today , and use that as a guide to see how much data you need to get between sites in a given replication cycle.

[albertwt]

As long as you are only going to replicate to that standby site , you wont need any additional licencing for it. You only need to licence the source for a backup or replication job.

you might not need WAN acceleration if you have sufficient bandwidth between your sites to replicate the workloads with a frequency that matches your required RPO.

If you have general purpose WAN accelerators in place like a riverbed you may find that unless they are specifically for Veeam backups , that they cache just fills up with Veeam data , displacing the data you actually want cached for general operations.

If you look at the size of your incremental's today , and use that as a guide to see how much data you need to get between sites in a given replication cycle.

Thanks Chris for the detailed reply.

So in summary, for the Storage infrastructure capacity required to do the above, it must be the same or larger size than in the production.

[dellock6]

Correct,
being replication you cannot leverage any data reduction as in backup. The final size at destination is going to be "at least" the one of the production VM, plus the additional space for snapshots used as retention points. If it's going to bi bigger than production however also depends if you are going to replicate every production VMs or a selection of them.

Luca

[albertwt]

Correct,
being replication you cannot leverage any data reduction as in backup. The final size at destination is going to be "at least" the one of the production VM, plus the additional space for snapshots used as retention points. If it's going to bi bigger than production however also depends if you are going to replicate every production VMs or a selection of them.

Luca

Many thanks for the clarification Luca. Previously I was looking for some similar DR VM replication equivalent with VMware SRM and EMC SAN based snapshot technology (Recover Point or MirrorView), but it is hard to configure and more expensive than using Veeam.

[dellock6]

Thanks!
Actually, now that I think, there is one data reduction you can use, that is to replicate a thick virtual machine into a thin one, this is perfectly possible and can save you some space. If instead you are already using thin virtual disks at source, the same size will be consumed at destination.

Luca

[albertwt]

Thanks!
Actually, now that I think, there is one data reduction you can use, that is to replicate a thick virtual machine into a thin one, this is perfectly possible and can save you some space. If instead you are already using thin virtual disks at source, the same size will be consumed at destination.

Luca

Ah yes, that does smart and make sense to replicate the Thick Production VM into Thin DR VM, and then after the DR testing we can delete the VM and start new again or in the event of real Disaster, we can commit back the changes to the Production VM.

is that possible with VBR 8.0 Enterprise license ?

[dellock6]

you do not need to delete any VM at DR site after tests, SureReplica can manage tests and replication at the same time. After tests what is removed is a protection snapshot, the replica VM is not touched.

For licenses, any edition has replication and complete failback capabilities.

[albertwt]

you do not need to delete any VM at DR site after tests, SureReplica can manage tests and replication at the same time. After tests what is removed is a protection snapshot, the replica VM is not touched.

For licenses, any edition has replication and complete failback capabilities.

Excellent, this is why I love Veeam software it is simple and works without having to invest in expensive and complex technologies like VMware SRM & SAN specific replication technology.

Thanks Luca