At Westminster College, we're at a backup crossroads because of the following reasons:
- We've been using Backup Exec for a while and we're out of licenses, but we need to deploy additional servers and services and be able to protect some new workloads, including Exchange 2010 and SharePoint 2010 data.
- We've started to push the limits of our backup window, making it difficult to contain backup operations in a time period during which they will have minimal performance impact.
- Our current backup method doesn't provide us much in the way of interim protection; we do daily differential backups in between weekly full backups, but we do not have the ability to easily implement more granular data protection methods.
- When recovery operations need to be performed, they can be monotonous, time-consuming tasks because we're still fully reliant on tape as our primary backup storage vehicle.
We're a very Microsoft-centric shop. Microsoft provides academic institutions with deep discounts that entice us to consider their solutions pretty carefully before moving on to other options. Without going into our selection process, I'll just cut to the chase: We're moving to Microsoft's Data Protection Manager 2010 (DPM 2010). DPM 2010 provides full support for every service we run, including:
- A mix of Windows Server 2003/R2 and Windows Server 2008/R2 operating systems.
- A mix of Exchange Server 2007 and Exchange Server 2010 servers, although the Exchange 2007 system will be decommissioned in a few weeks, as we complete our migration to Exchange 2010.
- A mix of SharePoint 2007 and SharePoint 2010 servers. Our institutional website and intranet uses SharePoint 2007. We're in the process of quickly moving the intranet services to SharePoint 2010 and will move the website sometime next year so, for now, we need to back up both services.
- SQL Server 2008 R2 servers. Our ERP (all of our software, really) runs on a SQL Server 2008 R2 system. We're working on implementing a failover cluster for this service.
Although we're heavily invested in VMware, we're going to maintain our server-centric backup processes rather than use newer virtual host-based options for two reasons: cost and complexity. DPM 2010 is extremely affordable and easy to use, and my staff and I are very familiar with server-based backup and haven't yet seen the business need to move to anything else. If you have a different opinion, I'd love hear about it in the discussion. Although DPM 2010 can back up Hyper-V based workloads, as I mentioned, we're a VMware shop with no plans to move away from that platform anytime soon.
We're also going to implement a disk-based backup method that sits in front of the tape library that we already have. This new disk system is a Dell MD1000 array with 30 TB of raw space (15 x2 TB drives). We'll deploy this as a massive RAID 6 array with one hot spare, so we'll be left with about 24 TB of usable storage (two disks will be dedicated to parity and one as a hot spare). We're backing up about 4.5 to 5 TB of data at present and expect that to grow to 5.5 to 6 TB in the next year. For DPM 2010, Microsoft recommends that the backup storage pool contain 1.5 to 3 times the total backup size. With 24 TB usable space, we will have plenty of room to grow.
On the server side, we've purchased a new Dell server with 8 GB of RAM as per Microsoft recommendations and a RAID adapter appropriate for the MD1000. We actually house the backup server, storage, and tape library outside the data center in another secure campus location.
Once we're on DPM 2010, we'll also schedule periodic snapshots to be sent to the DPM backup pool as a part of our backup strategy. For example, for our SQL Servers, I expect that we'll do twice daily differential backups in order to shorten the recovery window. For longer-term storage, we will continue to use tape. Our current tape library has a lot of capacity and is in good shape and will stay in service.
By moving to DPM 2010 with a disk-based intermediate backup storage location, we'll enable quicker restores and will have the ability to back data up more often during the day with a tool that makes the process relatively simple.
We recently had the opportunity to test our backup in a real-world scenario. Although we periodically test our backups by doing real restores, this was a little different. We needed to restore our ERP database from May 28th in order to meet a need that was not anticipated earlier by another office. I asked my DBA and network admin at around 1:00 P.M. to get it back and by 3:30 P.M., I received an email with the news that the database was awaiting my use. Sure, this is exactly what should happen, but it's always nice to get good news!
Keep up with Scott Lowe's posts on TechRepublic
Since 1994, Scott Lowe has been providing technology solutions to a variety of organizations. After spending 10 years in multiple CIO roles, Scott is now an independent consultant, blogger, author, owner of The 1610 Group, and a Senior IT Executive with CampusWorks, Inc. Scott is available for consulting, writing, and speaking engagements and can be reached at firstname.lastname@example.org.