Also tried to capture stats of eval method, some of the loop iterations of sequential subreader of index reader took more than 1 second to iterate over all the terms.
Tried with more RAM (120 GB) and loaded entire index into RAM via cat indexpath/* >> /dev/null but no improvement in query time. Also tried testing after integrating Lucene’s MMapDirectory into solr 1.0, still query time between 2 and 3 seconds
Any suggestions or pointers in the right directions will be very helpful.