If I can do it shouldn’t the Database be able to?
Back at the ZIPR I used to get teased about things like BJava (Ben’s Java) and BOracle (Ben’s Oracle) since I would often find annoying and limiting aspects of a Technology and then I would go on and on about what a POS they were.
Now I work a lot with MS SQL Server, which I enjoy working with much more than my old friend Oracle. So this is no way a reflection on one POS Database or another POS Database.
Anyway, if I can tune a query, shouldn’t the Database be able to do it for me? Why is the optimizer more often then not wrong? Why do I need to use join hints and join types? Why will it even allow me to add a redundant index? I’d really like to see some improvements in these areas. I feel so limited in what we can do with DSQL with a POS optimizer. Imagine a world where you write a query and the Database figures out how to make it run fast. I don’t care if it re-writes the query for me. Why should I have to worry about join order and indexing? Can’t the database figure out I need an index and what columns should participate in that index? Java has done a fairly good job with their Hot Spot Technology.
It is time for the Databases to shape up since there is no way they are going to slim down in todays world of unsurpassed data.
No comments yet
Leave a reply