Last night a datacenter saved my life, er… , app!

by Marc Watley on August 15, 2011

Alright, so you’ve toiled and slaved and spent countless sleepless nights (and probably countless ducats as well) to launch your shiny new mobile-local-social app that’s turned into the next big thing. Congrats and all. No really..no small feat, this; good work! Though before you pop open that magnum of La Grande Dame, allow me to ask: happy with your disaster recovery plan?

You. Do. Have. A. Disaster. Recovery. Plan. Right??

No? Step into my office. A Disaster Recovery plan is not the sort of thing you consider briefly, only to shove in the corner for a rainy day. All that’s needed is a cut of some errant fiber line, or some natural disaster to come along and your über-popular app could be down. As for you: that’d be time to dust off the ole resume once again. (I’m hearing the faintest sound of a teeny tiny violin playing.)

Let’s avoid this sad-sack scenario then, shall we? The good news is that although DR planning is a serious undertaking which should be performed by – ahem! – professionals, there are a few easy steps you can take right now to help appropriately prepare you for the unthinkable:

  • Ask yourself: What’s the worst that could happen? No, seriously. If that volcano erupts and hot liquid magma melts the datacenter where your servers live, it’s trouble. (Volcanoes do erupt, you know.) You’ll first need to determine how many hours or days your app can be down before your business is irreversibly affected. This is known as Maximum Tolerable Downtime, or MTD.
  • Research, research, research. Now that you know how quickly you’ll need to be up and running post-blackout, time to start looking into backup options. If your app is running on, say, Rackspace Cloud, you might want to start by learning about their load balancers-as-a-service, and reach out to your account manager for additional detail and customization. Similarly, if you’re running Amazon AWS, their Elastic Load Balancing offering is where I’d start. (With Amazon, you’ll almost certainly want to work with a company who has experience with tailoring  failover using AWS) Spend your time wisely and dig (and Digg) appropriately. Be super conservative as well – underestimate your MTD by perhaps 30% and ensure the solution you’ve picked can meet this. In writing.
  • Go time. Once you’ve selected your backup and recovery solution, time to implement and test. A weekend maintenance window is a good time to test and ensure that your DR system works. Make any adjustments and test again right away. Repeat this processed until perfect.

Keep in mind that his is the quick-and-dirty version of DR planning, but hopefully will give you a rough idea and perhaps serve as a primer regarding how you might approach this very important part of your infrastructure.

So if you’re still asking yourself, should I…?

The answer is simple: you bet your sweet little app!

Related Articles

Previous post:

Next post: