One of the patterns I've seen over the years is how IT and technology in general magnifies the consequences of poor performance. We put everything on one storage device so it's easy to manage, but it's also easy to type "drop table..." in the production database. It's also very hard to get the backup just right, and harder still to be able to test restores of a production backup.
As an IT manager, you need to mitigate for this. Do reviews, walkthroughs, or whatever you want to call it to ensure that at least two pairs of eyes have evaluated your designs. Do as much testing as possible. Fight for more budget for testing. Build the simplest possible solution that meets your needs.
Another thing you can do is subscribe to the comp.risks RSS feed and read all the different ways that the best laid plans can be derailed by one little mistake.