AnsweredAssumed Answered

Lucene indexing question

Question asked by steventux on Mar 27, 2009
Hi,
I've been through quite a bit of documentation trying to work out how to make lucene perform well in the following scenario. From what I can gather it's likely I am doing something or not doing something which means my application does not use lucene to the greatest effect.
I have approx 500 nodes/items created by a WCM XForm all in the namespace http://www.alfresco.org/alfresco/blog, I index properties like Location, Source and Author and for the most part a search by location is very quick, but a query just for blogs i.e. just
\\{http\\://www.alfresco.org/alfresco/blog\\}
is very slow. On a quad core 64bit RHEL server with 4GB Ram the query can take 12 seconds to retrieve 500 items. This timing is measured either side of the
store.luceneSearch('\\{http\\://www.alfresco.org/alfresco/blog\\}');
call.
A java backed script is quicker and of course provides more control over result set size and sorting and apparently after running once returns much quicker but the performance is still questionable. Am I missing a trick?
Should I be indexing a property just to identify the blog items in my repo or is the namespace query enough? I've also tried some of the default indexed properties but found no real magic there. I expect the web content growth to be significant and it worries me that I am currently seeing poor performance with so few records.
Any tips?

Outcomes