When The Daily WTF was still maintained it reserved an entire section of the site for Oracle Database. It may rarely have catastrophic failures (although it's the second one I've seen most catastrophic failures, just after MySQL), but it is composed of one unreasonable choice after the other.
Having worked at a company that used an Oracle cluster considering of over £1 million in hardware alone (I was never brave enough to ask about the licensing for the actual DB), the complexity of the setup far outweighs anything I could get my head around, and the company employed several full-time DBAs who were constantly tweaking parameters to keep the system performing - it could not simply be left to its own devices, there were humans in the loop around the clock.
Despite this, there was still a semi-major outage every few months that required even more complex failover systems surrounding it (message queues etc.) to deal with the downtime. I was with the company for only a few months but it was long enough for this to occur. Does make me wonder what the company was paying all this money (hardware, software and DBAs) for if the system was still expected to fail regularly.
