Veeam infrastructure

[ChrisJ83Knights]

Hi

We have just purchased Veeam B & R Enterprise Plus and are planning the infrastructure and seeing what hardware we will require or have available.

We will be backing up around 20 VM's locally and backing up (all) and replicating (some) to a DR site connected via MPLS.

I am going off the Off-Host Replication guide for Hyper-V in the Veeam Help Center, I believe we require a backup server, offhost backup proxy and a source wan accelerator at the main site. We will then require a target WAN accelerator at the DR site.

Is the above correct? Will we also need a backup server located at the DR site?

Spec wise can any of these boxes be Server 2012 R2 VM's or do any of them need to be a physical server? I can't find any reference to not using them as VM's in Veeam Backup Guide for Hyper-V.

Will we need to run SQL full for Veeam, can this be installed on the backup server? I saw reference to SQL full being needed for an environment with "a lot" of VM's but no reference to how many "a lot" is.

If VM's how many vCPU's, memory and networking would be required/recommended per role above?

Any help/advice appreciated.

[jmmarton]

Are all 20 VMs running on a single Hyper-V host? If so, then you could always use the on-host proxy. Also, for only 20 VMs, you could get by with SQL Express easily.

I'll let others comment on some of your other points.

Joe

[Mike Resseler]

I agree with Joe on the On-host proxy. In your case, I would invest my money in a decent server for the hypervisor (to make sure you have enough resources available both for the VM load as for the backup load).

You can install the entire backup server on a VM also, but if you have the possibility, I would advise you to buy a good specced backup server. That server then will be able to host all the roles for those 20 VMs on 1 box.

You don't need a backup server at the DR site, but you do need a repository over there. So you could create a VM at the DR site, add the WAN and repository role to it where the backup copy job will land the backup files.

Indeed. In this case no full SQL is necessary, the version included should be enough.

Hope it helps

[foggy]

And please see this user guide section for each component's system requirements.

[ChrisJ83Knights]

Thanks for the feedback guys, really appreciated.

Multiple hosts for the 20 VM's and to be honest it will be between 20 and 30 VM's.

When you say good specced could you advise roughly what that would be? I have the following options instantly available for instance, I can also upgrade the CPU's below if needed

24GB RAM, 2 x 4 Core CPU, Server 2012 R2, 7.5K disks

64GB RAM, 2 x 4 Core CPU, Server 2012 R2, 15K disks

128GB RAM, 2 x 8 Core CPU, Server 2012 R2, 15K disks

With regards to not having a backup server at the DR site what happens if the main site was to have a disaster of some sort and the Veeam backup server is offline, I would need to install Veeam in the DR to restore backups surely?

Many thanks

[jmmarton]

One question on the multiple hosts running the VMs. Are these multiple Hyper-V hosts in a Hyper-V cluster with shared storage or standalone Hyper-V hosts each with their own storage (or something else)?

Also, as to the DR site, it can't hurt to have a VM with Backup & Replication installed but not configured. That way in the event of a disaster you can easily restore the VBR configuration to that VM and then run restore activities. You could even go so far as to use a file copy job that automates copying the configuration backup from whatever local repository you use directly to the VBR VM in the DR site so that way it always has the most current config backup. But as Mike said, a VBR server is not required in the DR site.

Joe

[Mike Resseler]

I would go for a bigger specced Backup Server. I certainly would go for some faster disks because the longer it takes for the data to land at the backup storage, the longer your snapshot won't get commited on the hypervisor and therefore the longer you will have the resource penalty.

For the DR site. It depends on how fast you need to be. Joe's advise is good for faster recovery. You can open that VM and B&R, import the backups and start restoring. Otherwise you need to prep a VM first, install B&R and then import the backups. However, it does mean you need to get it up to date equally like the production B&R. Again, it all depends on how you organize your work and what is most important in your environment. Faster recovery (although we are talking only about 1 to 2 hours additional delay) or less time spend on administration

Hope it helps
Mike

[ChrisJ83Knights]

Thanks again guys and sorry for the delay in responding.

The hosts are standalone, so would 128GB and 2 x 8 core CPU's be overkill for a Veeam box running on-host backups?