It is a perfectly good Monday morning. The kids are off to school without drama (mostly), and I arrive at the office a few minutes early. The entire team is all smiles. Why wouldn’t they be? Someone brought donuts. Yes, indeed. The week is starting off pretty good.
I just sat down to continue working on a project that has kept me occupied for the past few months when I heard an explosion in the distance. Within a few seconds, all the power in the building was off. Completely dead. Nothing. As many of you are aware, stable electric power greatly enhances one’s ability to operate computer equipment. The sudden lack of power tends to cause problems, turning an otherwise great morning into something with the potential of going very bad.
But not in this case. Within a few seconds, the vibrations of the backup generators pulsed throughout the building. A few quick checks of the monitors and a visual inspection of the NOC indicated that all was well. Disaster averted.
Of course, our response didn’t happen by accident. We spend many man hours developing and implementing our disaster recovery plan to protect ourselves against this type of situation. And while we always hope for the best, we all know that eventually something bad will happen.
Most folks don’t need a plan sophisticated enough to handle a hosting operation, but everyone needs a DR plan of some type. And while you can spend hours obsessing over probability of occurrence and severity of impact calculations, a good DR plan is constructed around two very simple questions: what data will you need and how quickly will you need it?
The first question is the easiest. It’s a straightforward exercise to determine what data (or applications) you need. Most people already know. If you need a guideline, try this. Imagine that you just erased a bunch of data to make room for a new compilation of epic fail videos. Now, picture your spouse/boss asking you to print something from the data you just erased. How to envision that conversation going? There’s your answer.
The how-quickly-will-you-need-it question is trickier. Perhaps a better way to ask this question is, “How bad do you want it?” In general, a correlation exists between recovery time and expense — the faster you want it, the more you’re going to have to pay. If you can live without the data for a few days or a week, the $10 per month backup solution may be fine. If you need 5-9s, 24x7x365 access, be prepared to open up your checkbook.
The cloud provides numerous options for disaster recovery, many of which did not exist a few years ago. Many businesses, and certainly most home users, cannot afford to develop their own data centers. Moving data and applications to a resource that can realize the economies of scale that come from a shared infrastructure is a very smart move.
The cloud, however, does not solve all problems. This is best illustrated with a simple question — What if you can’t get to the internet? While the data may be safe, your staff of office workers is sitting idle. Or worse… your grandmama can’t see her pictures.
The bottom line is that disaster recovery is not a one-time event with a one-time solution. Preparation is a continuing process, gaining from the insight of each new experience. Like all critical response, disaster recovery should be choreographed and rehearsed. Otherwise, you’ll be left without a donut the next time the lights go out.