standardization on SQL was a key to achieving gorilla power.
In fact, my memory is that Oracle really pulled a fast one because IBM came out with the standard, making it single vendor by source initially, but offering it as a standard, but then did a terrible job of actually delivering. Oracle aced this by saying "sure, we'll take your standard" and then delivering. This was a lot of their early growth.
Row versus page level locking is not so much a client/server issue as it is a scalability issue. Certainly a part of Oracle's success has been being able to say "my database is bigger and faster than your database", thus not only attracting the top level, but also anyone below that who was concerned about growth.
Bingo. I have had precisely the same experiences. But the switching costs, in my opinion, have more to do with the non-SQL aspects of the RDBMS implementation than with SQL itself.
Indeed, in a good development environment, like Forte, for example, one can write the application so that it will run against any SQL database transparently, but one may gain better performance by noticing when the database is Oracle and supplying additional performance hints. |
