Dan Connolly's tinkering lab notebook

Adding server operations, DBA to my bag of tricks

I've always been a programmer. Then I learned enough about shipping software products at Convex and Dazel, and then global collaboration at W3C. I have always respected the people who keep the servers runningthe W3C systems team rocks!and now I'm learning a bit of that stuff too.
I mostly taught myself to program as a teenager. Then I worked at Convex, developing software supercomputers. It was a mature engineering organization; a great place to learn about shipping software products. Then I helped set up the engineering organization at a startup company, Dazel. Between these jobs, I learned the difference between a program and a product: a program works for the one who wrote it, while a product works for users. For the same set of features, developing a product is about a hundred times as much work as developing a program.

My job at W3C struck a nice balance between research and development for a good long while. In my short stint at Science Commons, I wrote "writing software to support research" in a status update, and I have hung on to that since. That's pretty much what I do at KUMC.

We're building a data repository for clinical researchers. It has just a handful of users now, but we're gearing up for more. So it's time to think about server operations. I did a little bit of system administration at W3C... it was something of a tradition that each new hire on the technical staff played that role... but eventually we grew up and hired real system administrators. Since then I have managed to avoid responsibility for server operations.

The informatics group isn't big enough for that yet, so I spent a few days learning the state of the art in open source monitoring tools. Getting ahead of the ball with service monitoring is a happy result.

It's certainly happier than the feedback I've been getting while trying to learn about Oracle administration. (Frank isn't the name this guy used, but he was very frank.)

if I make a cold backup on AIX (big-endian) and I try to restore it on linux (little endian), am I more likely to win or lose?
you need to use rman or [...]
but you guys really should hire a dba
you could afford oracle, and a server for aix, and now a new linux server
you can afford a dba for a few hours
Or else, you could totally screw stuff up, especially the configuration and tuning of the linux DB
most admins now are giant #$%@# idiots who dont know that sqlplus exists. 
or they are "developers"  which their company put them in charge of the database
and are giant clueless idiots