Another point is that with a b-tree you have your rows in sorted order, which means you can do "select * from blah where id > 100 and id < 200" and you're still

Another point is that with a b-tree you have your rows in sorted order, which means you can do "select * from blah where id > 100 and id < 200" and you're still using the index. If you were using a hash table, you'd be doing 100 random disk seeks instead of a few sequential ones.