Earlier this month, one of Microsoft's Azure data centers encountered an outage that impacted customers worldwide. Microsoft recently provided a preliminary report of the outage, which details the series of events that lead up to the outage. The outage of the storage servers impacted multiple Azure services such as Virtual Machines, Azure SQL Database, Backups, and so on. I'm sure that Microsoft's team handled the situation as best as they could have, incidents such as this remind us of the importance of having a recovery plan.
When it comes to SQL Server backups, one of the most followed rules is the 3-2-1 backup rule. For those of you are aren't familiar with the rule, it simply suggests you should:
- Have at least 3 copies of your data
- Store the copies on 2 different media
- Keep 1 backup copy offsite
Why should I have 3 copies?
Maintaining three copies of the same data may seem a bit excessive at first glance. From personal experience, I can tell you that it is the minimum number of copies that you will want to have for important data. In my case, I had several family photographs that I digitized when scanners became affordable. At the time, I thought I was playing it safe by saving those digitize photos on my computer, but also saving them to writable CD. Eventually, the hard drive on the computer faulted. The CD was also unreadable thanks to all the scratches that it accumulated over the years.
Why should I keep the copies on different media?
I think this is pretty evident in my photo scenario above. Had I only kept copies on my local computer's hard drive (perhaps on different drives), there's still a good chance that I could have lost both copies in one fatal swoop if I encountered something such as a power surge.
Why should I maintain a copy offsite?
Keeping a copy offsite doesn't seem like it would be necessary, but it's certainly another means of ensuring that the data could be available if something catastrophic should occur. For instance, I could have possibly burned my house down while attempting to cook dinner one day (I'm a notoriously bad cook). In doing so, I would have lost both my computer and the CD copy in a blaze of not-so-much-glory. I know that I've spoken to a few people here in Houston that noted that they were able to bounce back quickly during/after having their data center flooded thanks to the fact that they maintained copies of their data offsite.
The 3-2-1 rule is essentially about increasing the odds that you'll have a usable backup if the need for it should ever arise. If you haven't gotten around to using a similar rule, you may start considering it. While you're at it, you ought to also want to review your recovery plans and test them as well. I don't recall the exact phrase that was used, but it was something similar to "a recovery plan without testing is just recovery hope".