Re:Use SQL subqueries to count distinct 50x faster
by BigTuna on Sunday January 26, 2014 @10:06AM
(#7125275)
I've been rattling my cage about this problem for a few years now. There's room and need for an "ORM/DBA" position in our industry. Most of the common production packages are both complex and flawed enough to justify a full time position, and yet I've seen so many companies lose hours of productivity floor-wide because the dev group's [N]Hibernate knowledge has been pressed beyond its limits.
Even with simple CRUD I've had to unwind faulty implementations where a smallish (50GB) database was being eager fetched into memory because of poor ORM config and overuse of FK relationships - when a user did something as simple as logging into the app.
And I'm not talking junior developers - some of these guys were 10+ years at Amazon.