[ic] Which is best for performance/functionality mod_interchange or interchange under mod_perl

Mike Heins interchange-users@icdevgroup.org
Wed Oct 23 23:15:01 2002


Quoting Jeff Dafoe (jeff@badtz-maru.com):
>     While we are on the topic of performance, what are the thoughts on sql
> versus text when doing product scans?  I know on a well-memoried system (I
> have 2GB) that serious read caching occurs.  I also use postgresql, which
> has never been a real speed daemon.  This makes me think that using "text"
> for scans of the product database might actually occur faster than a sql
> scan.

Oh, for full-text it is in almost all cases. IC builds a screening
function which can do the equivalent of "grep" to eliminate lines which
can't match, and it precompiles the regex including checks for full-word
and case-sensitivity. This is very fast.

With the db search, we have to do a "select * from foo". That can be
a problem.....

If your database is less than a thousand records, either is fine. But
when you get above that much, full-text searching is much better done
with the text-based search.

If you get to larger tables, I built a couple of catalogs using Glimpse
for full-text search, one of which had 550,000 records and one with 1.3
million. The search time was hundreds of milliseconds for fairly-unique
terms, and this was when a 200 MHz Pentium was fast.

Also, for category searching don't forget the binary search of IC.
It would select categories from the above tables in hundreds of
milliseconds as well.

-- 
Mike Heins
Perusion -- Expert Interchange Consulting    http://www.perusion.com/
phone +1.513.523.7621      <mike@perusion.com>

For a successful technology, reality must take precedence over public
relations, for Nature cannot be fooled. -- Dick Feynman