Update on Stupid Database Tricks
Mar 14, 08:32 AM
Matthew Weigel was kind enough to point me to the full slides for the SQL Considered Harmful talk that I posted about yesterday.
Reading through the slides, I still want to barf when I see their “first we tried key-value pairs, then we tried context-less tables” on slides 19 and 20. Seeing them call slide 20 “good” is like hearing nails on a chalkboard.
They do admit that neither solution really worked and they now have their “Cryptic DB” solution, which apparently involves writing data to memory and then dumping it out to a DB at certain intervals, using a kind of journaling method. They somehow manage to not actually mention what the outcome of that method is; what does the data end up looking like in the actual database? Sure, it may perform well, but how the hell am I, the data analyst, going to get stuff out of the DB? I’m not saying their Cryptic DB solution is bad, I’m just saying that I don’t know enough about it to make that call.
