> From: Jonathan Mast [mailto:jhmast.develo...@gmail.com]
> Sure we can argue about which DBMS has the fastest JOINs but
> nonetheless it
> remains that JOIN queries will always be computationally
> expensive compared to single table queries.
Depends what you do with the results of those single table queries :-). I saw
one application (mid-1990s) that used SELECT * FROM table to bring two 10M row
tables into memory, then did O(n^2) comparisons between the keys to obtain data
from both. When asked, the developer said he didn't understand joins in the
database so did it "the easy way".
I'd hope developers are aware of computational complexity, load on different
components, and the reponse time and throughput constraints on the application,
and that they code appropriately rather than following dogma of the form "X
will always be expensive compared to Y". I've been surprised (rather too many
times) when the "will always be" has turned out to be false.
- Peter
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org
