Hard to say if the switch to Oracle (and the resulting expenses) was the reason for their demise, but I was surprised by this:

"In 2000 they were doing around $24 million per quarter with a gross margin of 21%."

Wow. How could you get off to a start like that and blow it all?

Well, it never was listed that Oracle was the reason, just that it was one of, and indicative of why it failed. If the MySQL solution was working, which the answerer on Quora implied, and there was no immediate sign of its failure, I think transitioning to Oracle at the cost of millions of dollars is a pretty big mistake.

Back then it might have been an OK decision, you did get some very nice things with Oracle including scaling and MySQL was a lot less mature, only about half a decade old. It certainly doesn't sound like choosing Oracle per se was instrumental in eToys' failure. On the other hand I've never heard a good thing about Sybase.

Sybase is a fantastic piece of engineering. Nice code syntax, rock solid in production.

There you go, now you've heard a good thing about the father of MS SQL Server.

Sybase has to be the worst of the "real databases", although SQL Server, DB2 and Oracle are bad too. The first time I tried to use Sybase, I wrote something like "CREATE TABLE foo (id INTEGER PRIMARY KEY, bar INTEGER NOT NULL)". Turns out Sybase doesn't have a type called "INTEGER". What the fuck.

Admittedly, all databases have their quirks. I use SQLite for a multi-process app suite, and get "database is locked" rather frequently. At least my app now knows how to retry transactions. Most of the apps using "real databases" fall over as soon as a transaction fails. "Internal server error, please retry later." Uh, no... transactions are supposed to fail. Even reads.

But I digress.

paul would argue otherise