Thoughts about Backup Strategy?

Question

We currently have 2 file servers in one location and one at a remote office. There's approximately 50Gb of data that is replicated via DFS and is also backed up daily to the disk and monthly to a REV drive. This strategy has proven to work pretty well. And in case one server fails catastrophically, the data will be available on two other servers immediately.

All 3 servers are also set as PDC, so in case one of them fails, there still should be 2 of them left standing.

However, the fact that OSs on the servers are not backed up, kind of bothers me. My plan is to get a relatively cheap(under 2k) 4Tb NAS and backup the whole servers weekly. Or does it make more sense to backup data and systems separately? What is your backup strategy?

Answer 1

I'm assuming that you're running Windows based on the fact you're running DFS. What I couldn't discern from your question was whether or not you're currently backing up the Active Directory and files to some other medium other than replication to the other sites. What is being backed up to the REV drive? My concern wouldn't be so much about the OS as it would be corruption of AD or a mission critical file that was then replicated out to the other servers.

I concur with your desire for a SAN for backups of the Servers but I'd be more interested in getting snapshots of AD and Files. Ideally you want to give yourself a few backups you can go back to in case of corruption or catastrophic failure.

In your scenario, I'd consider the following:

1. DC -> Backup AD to SAN.
2. Filestore -> Backup to SAN.
3. SAN creates snapshots daily (keep a couple of weeks worth depending on capacity)
4. Move a copy of backups offsite (REV drive)

As far as backing up the OS itself, there are a couple of considerations:

• What do you consider acceptable downtime for rebuild?
• Will you be rebuilding with identical hardware?
• Are you planning on storing the OS backup on the SAN?
• Level of acceptable loss (what is a backup worth to you?)
• What is the speed of your links?

Actual installation and configuration of Windows is a relatively quick process (in my opinion). Since the loss of a site is not likely to result in a data loss (due to replication) it's more of an issue of downtime. I'm not a huge fan of imaging servers unless there is a strong business case for it (ex: a time consuming and difficult configuration that will take significant time to rebuild). Images have their own sets of challenges and if you're going to add more admin time to a process, I'd make sure there isn't something that would fit your needs better first. Since you've got replication to other sites in remote locations, I'd estimate your biggest risk is unnoticed data corruption that gets replicated. Then you have to evaluate how difficult it would be to restore using your current method.

I personally don't see a lot of value in backing up the actual OS unless you're doing something on each box that would take significant time to regenerate. Target your files and the AD for extended backups and evaluate acceptable outage windows. Once you've done that you can better evaluate your DR needs.

Answer 2

Well, the first question is how much does downtime cost, and what scenarios are you looking at.

Then, look at how the restore process looks. If you figure that a serious loss probably means your hardware changes enough you would need to reinstall anyways, why back up the OS?

Personally, I like being able to survive the total loss of a location, but I'm paranoid.

Might want to say you are using windows (right? I'm a Unix guy) and what version, also, how much bandwidth you have between the sites, and can your users keep going if one server is down?