jackrabbit-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From AdamR <adam.r...@runbox.com>
Subject Re: Caching/performance issue (urgent help required please!)
Date Wed, 20 May 2009 11:10:01 GMT


Jukka Zitting wrote:
> 
> 
> What does the troublesome query look like?
> 
> BR,
> 
> Jukka Zitting
> 
> 

Hi,

The queries are fairly straight forward, for example:
/jcr:root/mynode//element(*, my:nodetype) order by @myProp ascending. I then
iterate through a hundred or so nodes from the resulting NodeIterator 
(which may contain many thousand results in total) and access various
properties and child nodes. It this latter bit that is extremely slow the
first time around. The slowness seems to be proportinal to the tital number
of nodes in my query results rather than the number of operations I perform
against the results. Subsequent runs are much faster, presumably due to
Jackrabbit's in-memory caching.

I've since read that XPath queries are not recommended for retrieving large
sets of nodes, and that programmatic traversal of the repository is much
faster. I'll refactor the code accordingly, but I can't believe that that's
the reason my app is taking minutes to load rather than seconds.

What's particualrly annoying is that that the during the first (non-cached)
run the CPU utilisation bumbles along at 10%. So what's causing the
bottleneck? MySQL doesn't seem to be running too hot and the only info I'm
getting from Jackrabbit are some messages re cache effectiveness.
 
-- 
View this message in context: http://www.nabble.com/Caching-performance-issue-%28urgent-help-required-please%21%29-tp23592190p23633390.html
Sent from the Jackrabbit - Users mailing list archive at Nabble.com.


Mime
View raw message