I have a pretty standard database schema here with the primary key prod_id
being my most often used join column in select queries. Categories, rankings
.. just about anything having to do with products uses prod_id in a join,
and user access on these queries is pretty heavy.
I wonder whether the performance of a select join depends on the length of
that column. Given that it is a primary key I have indexing on it, and all
other rules apply, but is there a difference in performance (CPU) between:
- 100000 products in both cases. One time the column is a MediumInt, the
other time a BigInt. I know there is a difference in disk space usage, but
is there also one in performance at all ?
- 100000 products in both cases. One table has its prod_id span from exactly
1 to 100000, the other one uses any random values between 1 and ... ahh lets
say 884570294. Any performance issue ?
Curious about what you guys have to say, anyway ...
Dan.
