hi all,
I use wampsever and mediawiki to structure local Wikipedia.I have imported the data of Wikipedia. At the very start,everything is fine. But a few days later,inquiry speed is slow.When I search for an article,it costs about 20 seconds.
Is there any way to slove this problem?
Thanks.
Katherine
IMHO WAMP server is a bad choice for the huge databases, because you need to optimize the parameters of MySQL (the my.cnf file) for the speed uploading of data.
See e.g. http://synarcher.sourceforge.net/#mysql_optim
Best regards, Andrew.
On Wed, Apr 21, 2010 at 5:56 PM, 李琴 qli@ica.stc.sh.cn wrote:
hi all,
I use wampsever and mediawiki to structure local Wikipedia.I have imported the data of Wikipedia. At the very start,everything is fine. But a few days later,inquiry speed is slow.When I search for an article,it costs about 20 seconds.
Is there any way to slove this problem?
Thanks.
Katherine _______________________________________________ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
Did you use the same search engine that Wikipedia uses? It's not the default MediaWiki search. I'm not sure if the squid servers help speed up search any, but there are huge amounts of hardware and dedicated to making Wikipedia run quickly.
-Courtney
On Wed, Apr 21, 2010 at 5:56 PM, 李琴 qli@ica.stc.sh.cn wrote:
hi all,
I use wampsever and mediawiki to structure local Wikipedia.I have imported the data of Wikipedia. At the very start,everything is fine. But a few days later,inquiry speed is slow.When I search for an article,it costs about 20 seconds.
Is there any way to slove this problem?
Thanks.
Katherine _______________________________________________ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l
On Mon, Apr 26, 2010 at 12:37 PM, Christensen, Courtney ChristensenC@battelle.org wrote:
I'm not sure if the squid servers help speed up search any
No, search is handled by Lucene. If you're using MyISAM fulltext search, it will be extremely slow on Wikipedia's data set, as well as having fewer features -- using Lucene is a much better idea.
wikitech-l@lists.wikimedia.org