Currently Katta, will pull the index out of HDFS and deploy it on local disk. HDFS is not known for low latency random access, and index lookups generally require low latency random access.
I seem to remember that someone had a patch that allowed lucene to directly access files in hdfs for the index, for readon ly access, but I do not remember the reference. On Sun, Oct 4, 2009 at 4:19 AM, Chandan Tamrakar < [email protected]> wrote: > I was using the hadoop.contrib.index code and was able to make a > distributed Lucene index > > Could then search over that index while it is still in hdfs > > > > Does the Katta project uses same hadoop index-contrib to index the > documents ? > > > > I found that there are also contributions solr-1395 and solr-1301 ( solr > -hadoop ) what would be the best approach to begin with > > > > > > > > Thanks > > > > > > -- Pro Hadoop, a book to guide you from beginner to hadoop mastery, http://www.amazon.com/dp/1430219424?tag=jewlerymall www.prohadoopbook.com a community for Hadoop Professionals
