Microsoft are asking Windows users to enlist family members to help kill off Windows XP as the support window for the aging OS closes on April 8th. This isn't the first time they've done this either, with a similar campaign to get rid of the less-than-beloved Internet Explorer 6. Anyone who has worked to retire a piece of software or business infrastructure knows that it can be a maddening, sprawling process that, even with the best planning, will always end up touching on more than you originally thought or intended. The scale of trying to retire a piece of commercial software with the install base Windows XP has, especially one that will still be business-critical for many people, is pretty much inconceivable to most of us.
Most DBAs who’ve held the job for more than a few years will be familiar with the upgrade cycle - there are always new versions and faster drives, better hardware, or the opportunity to offload sections of your infrastructure to a 3rd-party provider. All of the day-to-day planning, maintenance and documentation should in theory make this process as simple as possible, but in reality it tends to reveal technical debt - all the cruft, quick fixes and flat-out broken parts that we intended to fix just as soon as there was time. It’s also difficult to know the true impact of making changes - you might know every report run against your databases, but it’s much harder to know what business-critical processes rely on that data once it’s dumped into Excel and passed to someone else.
There’s a reason COBOL programmers can earn so much - the pain of digging out an entrenched system can be so great that it’s easier and cheaper to hire an expensive specialist than to replace ancient systems with something more widely-understood and easy to hire against.
The end goal for the DBA is the same as the people responsible for retiring products at Microsoft - happy customers (be they internal or external to the company) at a minimum of fuss. You can never account for everything when replacing and upgrading systems, but it’s one of those areas where foresight and planning can pay huge dividends. Still, next time you're cursing inadequate documentation or retiring a system whose dependencies seem to multiply like rabbits, spare a thought for the person whose job it is to get your grandparents off IE6.
