Anyone who has ever watched the MySQL query log while koha serves requests knows that koha makes A LOT of repetitive SQL queries for every request.  Most of these requests are for individual values in the systempreferences table.  There is also another large batch of queries for data from the language_descriptions table.  The queries are not that heavy by themselves, but there are way too many of them happening per request, and it’s for the same data every time.

In an attempt to minimize the number of repetitive queries, I tried to make koha use memcached.

First, let’s look at the problem we’re faced with:

All those queries are happening on every request.  The goal is to eliminate as many of those as we can using memcached to store the data once we retrieve it from MySQL.  That way, we don’t have to ask MySQL the same thing over and over again.

The patches that implement this can be found at:  http://arwen.metavore.com/~jsb/patches/memcached/ .

After those patches have been applied to the koha source, the number of queries made to MySQL are drastically reduced.

Unfortunately, preliminary benchmarks showed that this only improved the request rate by about 10%.  It was hoped that the speed increase would be more dramatic, but we’ll take what we can get.

Note that this set of patches has not (yet) been included into the main branch, but it may find itself there in the future.

The original announcement can be found at: http://lists.koha.org/pipermail/koha-patches/2009-February/002507.html.

Tags: Koha, memcached, Perl

This entry was posted on Monday, March 23rd, 2009 at 6:30 am by john.beppu and is filed under Koha. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.