Big Bang Doesn’t Work
Tuesday, April 13, 2010 at 5:00PM “We have no choice. We have to do a hot cut Friday night. It’ll be tough, but we’ll have the weekend to clean up. We’ll be fine on Monday.”
Ever hear words to this effect? I have. Multiple times. And I’ve always regretted not pushing back when I hear them! Whenever I’ve been lulled into the wisdom of the “big bang”, minor issues are amplified due to the scope.
Whether bravado, fact or simply be worn down, the charge to victory was compelling. While nobody likes hearing “Big Bang,” words like “simpler, cleaner, quicker, less risky, technically required” get tossed around like a football on a fall weekend.
My contention is people lull themselves into a sense of “Big Bang” being the only way to do a deployment. The ugly truth is with the exception perhaps of the Year 2000 a decade ago, most times a “big bang” isn’t needed. In fact, NOT doing “big bang” often requires more planning and creativity and yields a more satisfactory cutover.
For example, when deploying a new phone system, why not do a group or floor at a time? And can’t the old phones be kept in place for a week or so (if only for outgoing calls?)
When deploying a trading system, perhaps start with one commodity type.
Moving a data center? Bring up the network to the other center, and move servers/applications/data a little at a time.
Yes, you may have to run systems in parallel. You may have to have contingencies for “failing back.” Your users will have to understand how you are doing the transition, and will have to help make sure systemic controls are in place.
These are good things.
You will have to make sure you keep the books and records of the firm intact. There’s no excuse for values getting corrupt due to the transition.
If they worse thing happening during a transition is your users declare the target environment solid and want to accelerate the transitions…that’s a good thing. And this isn’t an invitation for “big bang” after a short trail….stick to your guns for an orderly transition as planned. Acceleration is good, having an excuse to do “big bang” isn’t!