We are implementing a pretty extensive system of database, data marts, and a data warehouse that track product and market research data. What I am finding is that extensive database systems are major pain. The application developers are frustrated because the frameworks are hard to use and their development environments are now cluttered with complicated database frameworks. The database developers are frustrated because stuff now breaks in new and mysterious ways. The sys admins are frustrated because the data recovery and disaster recovery plans just became terribly complicated. The best analogy I have to describe all this is that it is like flying a C5 Galaxy trying to get it to go Mach 1 through a hurricane with one pilot on the pedals, one on the stick, and another on the throttle. Oh and instead of bolts and welds it is all held together by super glue and duct tape.
Some sick part of me gets a kick out of this. I know that despite the pain we will get this plane welded up and put the bolts in and if we can get to Mach 1 in something this big then lookout because nobody can do that. As well, it is a rare feat when enterprise data projects are taken on they rarely start off smooth and easy. Oh and by the way, the great clicky and droppy examples out there that show how quickly you can implement an ETL process or use an object relational model, that just gets you on the plane.