Frequent Words / Performance-Benefits 5.1?
Posted by:
Jonny 5
Date: July 21, 2008 08:57PM
Debian-Server: P4/1.6ghz/4gb ram
Mysql: v5.0.32/key_buffer=2gb
Dear Forum,
I developed a MySql-based regional search-engine.
Total datasize is now about 2.5gb whereas fulltext indices fit in about 600m. I optimized a lot (some stemming/pre-processing) and performance is pretty good as long as it doesn't come to frequent words. In natural language mode as well as boolean mode. Query-time could drop to ~0.5-1s. Average time is below 0.2s.
Most default queries are in natural language mode. I read something about increased performance and new index-structures in version 5.16+ for boolean mode.
Does anybody know, if default mode (natural language mode) improved too?
The problem-border for me is at words matching more then about 15-20% of all rows. I can't put them in stopwords, neither could I use some extra grouping column together with 'match against' = bad performance too.
I know about Sphinx and even tried it and was very impressed. But I would have to change a lot of code and still hope to get performance increased with MySql ft-search.
Tia & greetings,
Robert from Austria
Subject
Views
Written By
Posted
Frequent Words / Performance-Benefits 5.1?
4590
July 21, 2008 08:57PM
Sorry, you can't reply to this topic. It has been closed.
Content reproduced on this site is the property of the respective copyright holders.
It is not reviewed in advance by Oracle and does not necessarily represent the opinion
of Oracle or any other party.