4 ms·
Thing is, there was a long period in the '90s when Oracle was the best solution, period. Non-mainframe DB2 was by the mid-90s (and perhaps earlier) nice, cheap
by hga 16y ago
Thing is, there was a long period in the '90s when Oracle was the best solution, period. Non-mainframe DB2 was by the mid-90s (and perhaps earlier) nice, cheaper and simpler to administer but used conventional locking instead of MVCC (works for a small group of DB engineers but doesn't scale as well with lots of developers), MySQL was a non-ACID nightmare (heck, unless my research was really off in mid-1997 Linux didn't truly impliment fsync back then), PostgresSQL wasn't ready for prime time and the less said about SQLServer the better (it's improved since them or so I'm sold). And small fry like Informix weren't particularly worth dealing with, Ingres was dying, etc. etc.
You can see how the above golfware paradigm could easily sink in when you really did have arguably the best solution, even if it cost twice as much as the next best (DB2, who's ecosystem (e.g. books) was and I'm sure still is much less rich).
However as you and many others have noted the world is very different now.
- tlack 16y agoI remember the same situation around those days: everyone wanted to get a database behind their site, but MySQL felt like you were running with scissors. It was easy to sell Oracle to clients then despite the $5k+ cost. Oracle's sales cycle may been protecting them until now, but they must realize what will happen when they abandon their strong technical underpinnings.