Minor clarification: The storage of indices uses the Hadoop file system API- not hdfs specifically - so connection is actually not to hdfs ... Solr can distribute indices for failover / reliability/ scaling to any hcfs compliant filesystem.
> On Jun 30, 2014, at 11:55 AM, Erick Erickson <erickerick...@gmail.com> wrote: > > Whoa! You're confusing a couple of things I think. > > The only real connection Solr <-> Hadoop _may_ > be that Solr can have its indexes stored on HDFS. > Well, you can also create map/reduce jobs that > will index the data via M/R and merge them > into a live index in Solr (assuming it's storing its > indexes there). > > But this question is very confused: > "Is this a better option for large data or better > to go ahead with tomcat or jetty server with solr." > > No matter what, you're still running Solr > in a tomcat or Jetty server. Hadoop has > nothing to do with that. Except, as I mentioned > earlier, the actual index _may_ be stored > on HDFS if you select the right directory > implementation in your solroconfig.xml file. > > So we need a better statement of what you're > trying to accomplish before anyone can say > much useful here. > > Best, > Erick > >> On Mon, Jun 30, 2014 at 2:19 AM, gurunath <gurunath....@ge.com> wrote: >> Hi, >> >> I want to setup solr in production, Initially the data set i am using is of >> small scale, the size of data will grow gradually. I have heard about using >> "*Big Data Work for Hadoop and Solr*", Is this a better option for large >> data or better to go ahead with tomcat or jetty server with solr. >> >> Thanks >> >> >> >> >> -- >> View this message in context: >> http://lucene.472066.n3.nabble.com/Integrating-solr-with-Hadoop-tp4144715.html >> Sent from the Solr - User mailing list archive at Nabble.com.