Josh Berkus wrote:
>
> David,
>
> > You don't think having a separate reference table for each code
> > lookup
> > -- that is, making the datamodel more fully normalized will not
> > impact
> > performance? Probably nominally because there will be so few rows in
> > them.
>
> It will impact performance no more than, and perhaps less than,
> evaluating a multi-stage CASE statement for each row in the query. And
> this kind of performance concern is only an issue if you're trying to
> run a public web site on budget hardware ... otherwise, the other
> elements of your system will be more of a bottleneck than the query
> parser! I regularly use queries and views that involve 6-9 tables,
> three UNIONS, and two sub-selects in each UNION ... and still get a 2-3
> second response time on the 500mhz Celeron production machine.
Thanks. Good advice.
Question: How big are those tables you're joining and uniting? (disk
size, and number of rows).