I went ahead and patched locally the SolrQueryParser in current 3_x branch.
Doing a quick test, baring any obvious mistake due to sleep
deprivation I get close to a 10X performance boost from 200qps to
2000qps.
I opened https://issues.apache.org/jira/browse/SOLR-2493
cheers,
-- stephane
On Tue
I' m using Solr 3.1 right now.
I was looking at a threadump trying to figure out why queries were not
exactly fast and noticed that it keeps parsing xml over and over from
the schema to get the lucene version.
SolrQueryParser are created for each request and in the constructor
there is a call sim
