DPM 2010 - Protecting and Recovering SharePoint, Exchange and Files

Since the last post I wrote on DPM 2007, our environment has undergone an number of upgrades:

SharePoint 2007 is now 2010.
Exchange 2007 is now 2010 (we waited until DPM 2010 was released so we could protect Exchange 2010 straight away)
SQL 2005 servers have been replaced by 2008.
All of our servers now run Windows 2008 R2.

To ensure all the data we have in those new systems is protected, we had to upgrade our DPM 2007 server to 2010.

In-Place Upgrade - Not So Good

In an effort to retain disk-based recovery points, I took the (wrong) decision to do an in-place upgrade of our DPM 2007 server, which was running on Windows 2003 R2 x64. As DPM 2010 requires Windows 2008, the first job was to upgrade the operating system, which went comparitively smoothly, and DPM 2007 continued to function after the upgrade - so far, so good.

However, the second upgrade, from DPM 2007 to 2010, wasn't so smooth. Although this is a supported scenario, and DPM setup will walk you through the upgrade, something went wrong with ours. Everything looked OK - we were left with a working console, and the protection groups were all present and correct - but nothing was backing up, even with the updated agents. We raised a call with Microsoft PSS to help (it was a production system after all) but a after a few days of sending massive log files backward and forwards, decided the best way was just to run a fresh install of DPM 2010 and recreate the protection groups - must of the disk recovery points had already expired anyway, and would import the tapes. Perhaps if we just relied on disk-based protection, we would have pesevered, but that's the advantage of good old tape. We'll put the failed upgrade down to one upgrade too far...

Much Better

Once up and running, DPM 2010 is well worth investing in. Its so much more polished than 2007, whilst still retaining the simplicity of the original. It's also far more reliable. I remember when we first started using 2007 in its RTM release, and needed almost constant attention to keep it up and running. Several service packs and hotfixes improved things a lot, but there would usually be at least one or two problems a week that required intervention.

Not so with 2010. Straight out of the box it just works. And looks after itself - if you let it, Consistency checks and recovery points running out of space are a thing of the past. And my personal bugbear with the old version, having to have individual space allocated to each protected SQL database - has now gone. We have many small databases that all need protecting, and before each one needed a few GBs each for recovery points in DPM (the minimum allocation being 1.1GB), even when they only used a few MBs of storage. Now, SQL databases share DPM disk space, using it far more efficiently for many small databases.

So DPM itself is a lot better - as well as the improvements in managing protection groups, the UI itself is a lot quicker, and it seems to handle more protected resources than the old one did on the same hardware. But the main reason we upgraded was to protect the new versions of its key workloads, SharePoint 2010 and Exchange 2010.

New Stuff

First SharePoint: With 2007, you had to use a recovery farm, running alongside your production farm, to recover files/sites back to the production farm. This meant you could recover just about anything easily, but maintaining that second farm was a pain, particularly if you only did restores once in a blue moon. Now (with DPM 2010 protecting SharePoint 2010) it's possible to restore using an "offline" content database - meaning you don't need a recovery farm. DPM just attaches and mounts the recovered database to your SQL server and SharePoint can read the DB without associating it to a SharePoint web application. Much better. Other than that, it's the same process as before for protecting SharePoint - you need to run configuresharepoint.exe on a farm server with the DPM agent etc.

Secondly, Exchange: Not much different here, other than DPM 2010 obviously supports the new Database groups avaialble in Exchange 2010. We've done a few restores now, and whether its individual items or an entire mailbox, its a doddle. Not had to do a full storage group recovery yet, but with DAG and DPM, I don't think that's ever likely without a full-on asteroid strike.

System Center Integration

Integration with System Center - both Operations Manager and Service Manager - is also a lot better now with the improved management pack that can use SLAs: instead of notifying about each recovery point failure (which with a few thousand a day going on can be irritating) - it only raises an alert (and therefore a SCSM incident) when the failures occur several times in a row.