The Certified DBA
thedailywtf.com
thedailywtf.com
Politically, you can run into weird "ownership" issues (eg, the DBA team handles system performance) and fighting the chain is soul-deadening. One of the reasons I considered actually getting one of the myriad DBA certifications is to add the heft of being able to say I had it to the PHBs when this type of thing comes up.
Thankfully, I held off and I think for the most part these certifications (although valuable when paired with demonstrable applied skills in real projects) have fallen out of favor lately. Certs are a particularly strange form of resume fodder - a few years ago they were pretty key in making it past the HR gauntlet. I imagine that for some techs (I'm looking at you, Oracle) they are still a defacto requirement.
...not to mention a career-limiting move. PHBs tend to not like it when people complain to them about their bad hiring decisions.
This story is actually a minor example of politics in play. If you're an idiot when it comes to corporate politics try and stay that way. Ignorance is bliss in this regard.
Yes, there are "Certified DBA's" who are morons. Guess what, there are System Admins that are morons, Programmers that are morons, Pilots that are morons, Doctors that are morons, etc, etc.
If his main takeaway from that situation was that "Certified DBA's" are morons and that system tuning is little more than hand waving and hot air, he failed to learn the real lesson.
The real lesson imo, is that in every field you have people that are morons, and you don't know if the person you are dealing with is a moron based on their credentials. You only know if someone is smart and gets things done through actual experience.
For what it's worth, performance recommendations of this sort are the easiest to expose because of the availability of quantitative performance measures.
Luckily, there are great tools out there like ORION (Oracle I/O Numbers Tool) that let me simulate an Oracle workload which give me real quantitative data that I can use to prove a particular configuration is non-optimal.
I've had to prove and disprove configurations before, and personally I think that if you're building any storage for database, you better benchmark it. Too many sysadmins just take the default settings on their storage and might only get 10% of the potential out of it.
I've seen a lot of systems that were setup in wtf ways. For example, having 4 HBAs in a system and all data is flowing down a single path, since the sysadmin never bothered to install multipath drivers or set path preference. This happens _all_ the time.
Check out my blog for some of my latest database storage benchmarking: http://sanarchitect.blogspot.com