High Level Target and Configuration Question

[philrich]

I just run the business - relying on expertise from my staff - but I wanted to vet' this before it went any further...

My environment consists of a three server HyperV 2008 cluster (three Dell r910s) that host approximately 24 VMs (all Win2008 - and include two instances of SQL 2008 and one of Exchange 2008). Exchange is moving to Office365 shortly - so that's off the table. The SQL servers have ~2TB of data -- and the rest is misc. data files. All of this sits on a 18TB SAN.

Internally - everything has a 10GB backbone.

I have a spare Dell R520 with about 20TB of internal shortage (right now with Win2008) and a 4 foot high stack of Proliant DL380's and DL360's that we decommissioned.

Because of compliance requirements -- the goal is to back everything up locally to disk - transfer a copy off site - and get myself in position to spin up the backed up environment in a FISMA moderate accredited environment (right now Amazon's FedCloud)

As I understand it - the plan is..

1) There's a target repository (the Dell R520) - but from reading here we should consider upgrading to Win2012 and turning on de-duplication.
2) The backup software would sit on the same machine as the target repository
3) We have Veeam kickoff our DBA approved backup scripts and back that up to disk - but that's really a independent backup done for his benefit that's not included in the Veeam backups.

Questions:

a) Where am I wrong on this high level plan?
b) Should the backup software sit on the hypervisor instead?
c) Given that, in the event of the building burning down, I need to spin up the VMs to FedCloud, thoughts on where I should transfer the offsite backup to?
d) Any use for the pile of decommissioned servers?

Thanks,

[PTide]

Hi Philrich,

1) There's a target repository (the Dell R520) - but from reading here we should consider upgrading to Win2012 and turning on de-duplication.

Correct, deduplication perfectly fits your case - a stack of 24 VMs, all carrying 2008R2 inside.

a) I don't know - can't see all the details from here =)

b) In case your repo+backup server suddenly dies both repo and backup server get lost, that's no good.

Personally I think that it'd be better to deploy your backup software on a separate VM inside your cluster and perform a VBR configuration backup on a flash drive or something + make a replica or a periodic full backup of your backup server. In case your cluster dies, you will be able to restore your backup server faster either by installing a new instance of VBR and deploying your conf.backup or just by deploying your backup_server_vm from full backup.

So, the answer is "yes" - the backup software should sit on the hypervisor.

c) It depends on the amount of resources you posess and on how paranoid you are =D

In my opinion there are couple "must have"s for the offsite:

- 365/24/7 accessibility
- it must be really offsite - next building down the street is not an option
- all systems independent from your production - Web, power etc.
- Somewhere 5-10 km away.

d) It depends on the reason why you decommissioned them...if they are still good, I'd make another hyper-v cluster of them and let it stay somewhere offsite in a power-down state. If things go really bad - you turn them on and your life gets a little bit easier for a while.

Hope this helps!