VMware Cloud Community
ctfoster
Expert
Expert

vSphere and Data Protection Manager 2007

Hi.

Anybody out there had any experience using Microsoft Data Protection Manager 2007 with VM's. The particular scenario I'm looking at the moment is using a second DPM server in an offsite location to backup the first DPM and then using this a a source for restoring to a set of VM's at that site.

Using one DPM to backup another for DR purposes is recommended practice - but I wonder if anybody has taken it a step further and used the backups on the remote site to keep a step of VM's in step through periodic restores. Before I try messing with this it would be great to get some feedback from somebody else who has tried it.

Cheers

CTF

Tags (2)
Reply
0 Kudos
8 Replies
ctfoster
Expert
Expert

Obviously not..oh well here goes!

Reply
0 Kudos
VMmatty
Virtuoso
Virtuoso

I'm curious how your experience is with DPM 2007, so post back when you have some results. Microsoft recommends using this product to backup Hyper-V enviroments, so the functionality is there and should probably give you what you want. The integration between DPM and Hyper-V is obviously much better than with vSphere, but the concepts are the same.

What DPM likely won't give you is the whole host style backup/restore capabilities that you'd get if you were backing up Hyper-V. You'd get that with vSphere if you used Veeam Backup or vRanger, but DPM doesn't (and probably will never) support vSphere directly.

Good luck, let us know how it turns out.

Matt | http://www.thelowercasew.com | @mattliebowitz
s1xth
VMware Employee
VMware Employee

I use DPM 2007 to backup VM's folders/volumes that contain important information with no issues. I would be interested to hear what your setup is and what you are trying to do with the VM's. If you are talking about just backing up the files/folders of VM's to say the DPM server than you should have no problems.

http://www.virtualizationimpact.com http://www.handsonvirtualization.com Twitter: @jfranconi
ctfoster
Expert
Expert

Thanks for the input.

The plan is not so much as to use DPM to backup/restore the VM's at a VMDK level but to provide DR cover for Exchange and SQLServer systems running on VM's. We've found Microsoft DPM pretty reliable backing up 'Volume Shadow Copies' of these transactional systems as well as standard 'file share' servers. By taking VSC snaps every hour, backup size can be kept fairly managable and after replicating them to an offsite location through a DPM->DPM backup it sees an obvious step to try and restore them to VM replica's at that site.

Clearly we are going to be behind the business by the time we have done all this but so long as everybody buys in and the expectations are managed this is not an issue and for the sake of giving it a go the customer gets automated offsite backups, a recovery plan thats tested daily and a poors mans DR.

The idea of using a two stage replication is that we can still restore and backup locally and it all runs at top speed - the backups are never throttled back by WAN issues. We will run the initial backup DPM->DPM locally before moving to the remote site.

All the backup traffic runs across a management LAN so its not slowing anything down. The other major plus is that the DR sites runs with the same address space as production - so I know there's a fair chance its going to work if we get data across. I dont have to run a whole load of scripts.

We have an idea of passing VMDK delta's of non-transactional server systems across the link using the same idea or using esXpress.It all depends on the spare bandwidth and the latency on the line. We can run them at night when the transaction stuff is quiet. A large number of the servers are just farm members either for citrix for web applications, they dont have to be that concurrent to get the site back up.

Never a dull moment.

Reply
0 Kudos
s1xth
VMware Employee
VMware Employee

Awesome. I am using DPM for the same exact setup acutally, although I do not have machines in 'standby' at my DR location 'yet'. That is my next step, how are you preparing to do the restore of these VM's? Are you going to have a running copy of the VM's running at the DR site and then just restore from the DPM servers the most recent data in the event of an disaster? How are you testing your exchange restore? I am using DPM right now on an Exchange vm and the only thing I do not like about it is that you have to restore to the same server that is running, you cant restore a db to another to totally different server. Just curious how you are testing the exchange restore.

Thanks!

http://www.virtualizationimpact.com http://www.handsonvirtualization.com Twitter: @jfranconi
Reply
0 Kudos
ctfoster
Expert
Expert

Are you going to have a running copy of the VM's running at the DR site and then just restore from the DPM servers the most recent data in the event of an disaster?

I guess thats the idea for the transaction servers such as SQLand Exchange - but try and keep the restores as up to date as possible. For instance we snap every hour on the production site and then backup the DPM server every 4 hours. Once we get these backups over we can try and restore these snaps. The servers on the remote site are a 'clone' of the production site.

In theory the situation is the same as if we were trying to recover from a failed DPM server which is a supported setup from Microsoft. So I guess thats a pretty good test. The trick is to get each DPM working with a couple of interfaces - one which links the DPM's and the other that the production LAN hangs off on both sites-both running the same subnet. Both sites know nothing of each others Production network. For this reason both copies Exchange servers can run active - the DR version constantly running restores on a schedule.

Well thats the plan - I stress we dont have the DPM on the DR running yet. This is all theory - which was the purpose of the post. Thanks for your interest.

Reply
0 Kudos
IRIX201110141
Champion
Champion

Well, i found most of the ideas here very confusing :smileymischief:

  • DPM always needs an Agent within the Machine you would like to protect

  • Protection for Volumes(Files), Systemstates and application like Exchange/MS SQL and Sharepoint are possible

  • Every Server have to be a member of a AD. The DPM it self and ever machine you would like to protect

  • Restore to other servers is possible but also have to be a member in your AD

If we're in VMware Land the DPM never sees a VM so a backup for DR is not possible because you need a working OS and Agent first. This might be different if you have a Hyper-V server because the Agent have access to the Vdisk files.

A restore of MS Exchange goes always into a Exchange Recovery Group. This can be testet with a running live system because mounting the RG is a manual step and the mailbox can be attached to a existing one if needed.

But comming back to VMware and a DR site i cant see how DPM can helps here because the server have to be a active AD member for doing a restore.Two server with the same SID/Name/$whatever never can be exists to the same time and i cant see how to hide the server from your network.

Regards

Joerg

'Remember if you found this or others answers helpful do not forget to award points by marking an answer as helpful or correct'

Reply
0 Kudos
ctfoster
Expert
Expert

Joerg.

I'm sorry if you found my description confusing. My initial post was just to see if anybody had done something similiar, it wasn't intended as detailed description of what we intend to do...which only exists on paper. (Although the local DPM is up and running)

First, I'm not sure that can only restore to a recovery group. That's certainly true if you need to recover a mailbox - but I'm looking to restore the entire storage group so I'll be using the option to recover to the original location as documented below. You would use this on a corrupt SG - which is what I assume I have.

I know the AD stuff sounds a bit wierd but look at it this way. Say you had a really bad day and you lost your whole Production site. All you have left is image backups of all your VM's vmdk's on a disc somewhere and a copy of your last Exchange/SQL backups on a DPM on a remote site. Could you get working again - sure.

Over the weekend you could rebuild your ESX, get some storage sorted out and restore your vmdk's and start them up. Once your Exchange server was back online - point it to the recovery DPM (Microsoft provide a script for that) and restore the storage groups back to the original locations. At least you have something that was as good as the last replicated backup set. Thats sort of what we are doing... except we are expecting the disaster!

The two AD production networks can't see each other so they don't get confused. All the DR site sees is another DPM server on a isolated LAN segment somewhere which has some data that it backs up. As far as the DR DPM is concerned the same (VM cloned) Exchange server that wrote the original backups periodically asks to restore from those backups overwriting any existing storage group.

Keeping AD up to date is a little tricker. You could use the same model (using DPM) but restoring AD in anyway shape or form to any past position is not my idea of fun. The only good copy of AD is the one that was running a minute before the rack caught fire.

What we plan is to have a 'tin' DC sitting on the management LAN at the DR site doing nothing but taking standard AD/ DNS replication from one of the VM DC's at the production site. W'ell have to punch a hole in the firewall and add some routing to allow this action to occur.

Should we lose the Production site we'll fire up the DR Vm's - which may be a few hours/ days out, and allow them to chat with the up to date version on the Management LAN before bringing everything else up.

I cant believe theres not a few gotcha's in this somewhere - but thats half the fun.

Reply
0 Kudos