particularly if you are linking requests. The latest RT code hopefully has fixes as a result of this work.
Dave
Steinar H. Gunderson wrote:
On Wed, Nov 24, 2004 at 09:57:52AM -0500, Christian Fowler wrote:
As for performance, lots of others will probably volunteer tips and techniques. In my experience, properly written and tuned applications will show only minor speed differences. I have seen several open-source apps that "support postgres" but are not well tested on it. Query optimization can cause orders of magnitude performance differences.
Definitely. My favourite is Request Tracker (we use 2.x, although 3.x is the latest version), which used something like 5-600 queries (all seqscans since the database schema only had an ordinary index on the varchar fields in question, and the queries were automatically searching on LOWER(field) to emulate MySQL's case-insensitivity on varchar fields) for _every_ page shown. Needless to say, the web interface was dog slow -- some index manipulation and a few bugfixes (they had some kind of cache layer which would eliminate 98% of the queries, but for some reason was broken for non-MySQL databases) later, and we were down to 3-4 index scans, a few orders of magnitude faster. :-)
/* Steinar */
-- Dave Cramer http://www.postgresintl.com 519 939 0336 ICQ#14675561
---------------------------(end of broadcast)--------------------------- TIP 4: Don't 'kill -9' the postmaster
