Hi Rui, You don't need to merge the resulting indices (1 index per Reducer is what I assume you are asking about). Each could be copied to a different Solr server and you could then use regular old Solr distributed search to search across them.
You don't want to search indices while they are in HDFS. That's why above I mention copying them. Otis -- Search Analytics - http://sematext.com/search-analytics/index.html Performance Monitoring - http://sematext.com/spm/index.html On Fri, Oct 12, 2012 at 5:47 PM, Rui Vaz <rui....@gmail.com> wrote: > Thank you very much for your replies, > > Yes Otis one possibility is to copy my data do HDFS and then apply a Map > function > to create the intermediate indexes across the cluster using SOLR java > library in HDFS. > > I have some doubts concerning this solution: > > 1 - The intermediate indexes created realy > need to be merged? > I mean is there any mechanis in SOLR > CLOUD to easily > combine those intermediate indexes > and serve them as if they were a "whole index", in a > distributed fashion? > > 2- Can I serve these diferent index with > SOLR or SOLR CLOUD directly in HDFS? > Google says no :), so maybe I need > to copy the indexes to a local file system > and point Solr at it. > > Timothy thak you for your tips. I am looking at Pig. > CloudSolrServer seams an interesting piece of the architecture, especially > to discover > Solr endpoints and then possibly replicate my index, but I was wondering if > I need to implement that > or if Solr will take care of that for me. Maybe I just didn't get your tip > due to my newbie knowledge in Solr. > > I am sorry if I am confusing some concepts or not being very precise in my > words. > > Jack thank you for sharing DataSax solution, I will definetey take a look > since it's free :). But anyway the objective > of this project is for me to learn Solr and Hadoop. :) > > Thank you, > Rui Vaz
