Disaster Recovery and backup solutions for VMware servers

[gnicklaw]

To all,

I am knew to this area and have been assigned the task to evaluate Disaster Recovery and Backup solution for our VM servers. I have been looking at ComVault, Acronis,  Veeam and  PHD. Does anyone have any suggestions?

Here is what we are looking for:

1. Ease of scheduling backups
2. Ease of restore of a server from backup to current location on to another Host
3. Restore to like on unlike servers, bare metal restore
4. Can the software rebuild a HOST, not just VM’s
5. We currently have CentOS, Ubuntu,Fedora, Windows servers, VMware ESXI and ESX, vSphere 3.5 and 4.0 going to 4.1
6. Speed of restore of server service on critical servers (Cold server ready on another HOST ready to turn up)
7. Disaster recovery
   1. In the event we loose a location how fast to bring up another location with the servers that are lost. (This will be in a different state)
   2. Need to backup SAN storage at one location to another location
   3. Reports and graphs  showing status of backups, backup schedule, time required for backups, etc
   4. Cost per HOST
   5. Required servers to run the backup solution
   6. How to have a 2^nd backup manager ready to take over in the event we loose the backup manager, HOT or COLD spair.
   7. Backup manager with redundancy

Thanks

GJN

[chriswahl]

Hi GJN,

I use the Veeam product for backups and Double-Take for replication to DR, as both accomplish different goals.

Regarding items #1 and #2, both are easy to schedule and offer a variety of options (backup a VM, folder, datastore, host, etc.), however I mainly use Veeam to provide local and tape restores, whereas Double-Take provides the high availability at a co-lo should my primary site fail.

I'm not sure how relevant #3 is, as the VMware hypervisor doesn't really care too much what hardware it's running on. It's up to you to purchase servers that meet the requirements of your VMs, such as featuresets on the CPU.

Veeam has a "spare tire" like restore, where you can power a VM up directly from a backup file without having to restore it, which is nice in the event of a critical failure that isn't solved by HA/FT (such as a corruption).

For most of the items in the #7 DR requirements, it depends heavily on what your SAN is capable of doing (example: hardware replication like an EMC CX vs requiring additional hardware or software replication as with an EMC AX)., what your consolidation ratio is per host, etc. I do know that Veeam can have multiple backup managers pooled together using their Enterprise management tool, offers reporting, shows status, and all that jazz.

It would probably be best to contact them for a demo and cost analysis.

Hope this assists,

- Chris

[gnicklaw]

Thanks for the information. I will look into the software you use also to see how they compare with the others.

Thanks again for the response

GJN

[Virtugirl]

With regards to the DR scenario, If you have your SAN replicating to a remote site, then you can use VMWare Site Recovery Manager to script the failover of your site.  Ive used it before and its a great product.  If you need anymore information give me a shout.

[gnicklaw]

What do you use for the SAN replication?

[Virtugirl]

If you have 2 SAN devices by a storage provider - eg NetApp.  One SAN device will be located at your primary site and one located at your recovery site.  The NetApp SAN will automatically replicate to a DR site, using schedules set by the Admin.  In the NetApp world this is called Snapmirror but other SAN technologies will have a similar product.

[gnicklaw]

Thanks

We use Overland SAN and PAC Data SAN's in our environment. I need to look for what will work with our devices.

Thanks for the feedback.

[chriswahl]

If your equipment does not support replication, you can still use a software replication strategy such as Double-Take. There are many offerings out there, but the key features to look for are WAN compression, transmission deduplication (to avoid sending any thing other than the delta changes), and proper failback handling (failover is often easier than failback).

One caveat I've found with hardware (SAN) replication is that it's ignorant to the vmfs datastores. If you do a storage vMotion from a LUN with a replicated datastore to a LUN that is not replicated, you'll lose that virtual machine at your DR site. Proper documentation and change controls processes are a must.

[Virtugirl]

Totally agree with Chris.  If you do have replication technology at the hardware level and support on your devices, it may be worth speaking to your SAN supplier to find out what options are available to you.

[gnicklaw]

Thanks

I will be checking with the manufacturer today and already requested a meeting with Double-Take also.

I have set up trial Tests now with:

PHD

Acronis

ComVault

now Veeam and Double-Take

Thanks for the information again