Thoughts on backups and RAID

I tend to rely heavily on my RAID 1 mirror for keeping backups of my files (supplemented by occasional DVD burns). I've also recommended RAID 1 highly to other people, especially for the type of folks that I knew wouldn't backup their files regularly. However, I recently made an editing mistake that almost cost me some data, even with my RAID 1 backup. Matt Jadud's recent post on backing things up got me to thinking about the strengths and weaknesses of my different backup strategies.

Here are my current thoughts:

RAID

• Mirroring prevents downtime due to hard drive errors or crashes
• if one drive crashes, the backup drive can be used until a repair is made
• new spare drives can be synched easily with the backup drive to get back up and running in minutes (you should always test this feature when setting up your mirror the first time!)
• Mirroring is automatic and seamless.
  
• With a hardware RAID controller, keeping your drives mirrored causes no discernable performance hit. (At least, not that I've ever noticed.)
• Unfortunately, with hardware RAID, you also may be tied to a particular controller card's data format. (Or even worse, you may be tied to a particular motherboard in case of onboard RAID.)
  
• If you have a controller card or motherboard failure, and your RAID controller is no longer widely available, you may be hard-pressed to retrieve your data.
  
• RAID 1 does not allow you to revert to a previous backup state.
• Once you've made a change to the data on your hard drive, the second drive has the exact same copy of the change.
• This means if you change or blow away a file by accident, RAID 1 by itself won't help you. Archiving your data in some other fashion becomes essential if you need a previous version.
• Mirroring is obviously somewhat more expensive since hard drives cost more than DVDs or tape. But this is becoming less of a concern as HD prices continue to fall.
• Mirrored drives cannot be stored in another location ... a big disadvantage in the case of fire, flood, lightning strikes, etc.
  

Media backups (CD, DVD, tape)

• Backing up to media helps you (to an extent) when your hard drive fails.
• However, you may not be able to get back up and running very quickly ... installation of OS, applications, drivers and other software is separate from restoring your backup. In my experience, you'd better count on at least two days downtime if you're rebuilding your configuration from scratch.
  
• Media backups are a manual process.
• You may be able to automate it to an extent, but you still have to switch out tapes, or DVDs at some point.
• Backups may cause a significant performance hit.
• With data compression turned on, your backups may bog down quite a bit, making it tough to keep using other applications.
• You may be tied to a particular type of media format.
• Tapes are the worst for this. Many DAT drives read and write to tapes in a specialized format. Good luck trying to find that brand again if it's been more than a few years since you bought the drive.
• CDs and DVDs are much better for this, although you still have to watch that the backup software you use doesn't change its data format either.
• Media backups allow you to restore to a previous backup state.
• This is, without a doubt, the single biggest advantage of backing up to media. If you accidently delete or change a file, retrieve the archive, and you're up and running with (hopefully) very little data loss -- depending on how good you are at regular backups.
• Media backups are very inexpensive.
• Media can be stored at a separate location to provide even more security.
  

I think I managed to convince myself that the approaches are complementary, and, in my experience, I now try to use both mirroring and regular backups to provide a good sense of security. Occasionally, I take my backup DVDs to work and store them there in case something ever happened at home. Sometimes that practice has paid off too (a few times where I needed a very old version of a file on a DVD that wasn't on the regular backup rotation).

A final comment: I also discovered a hole that neither mirroring or backups will catch unless you're aware of it. That area is storing files on the Web. I use Flickr for archiving some of my photos and Bloglines for reading RSS feeds. The other day, I accidently deleted 3/4 of my Bloglines feedlist by mistake. To my horror, I realized there was no way to undo my mistake and recreating the exact list would be almost impossible. Fortunately, I still have most of my feeds on another Web-based program called BlogBridge, but that was simply good luck.

The experience reminded me of all the online Web services we trust not to lose our photos, emails, address books, etc. This trust is simply misplaced. In truth, we can't count on those services to be reliable either (or immune to our simple mistakes!) Our online data should be as much a part of our backups as the data on our hard drives.