Nayan:
Check out the DSE 2.1 documentation regarding the "DSE Search Resource REST API". Unlike plain old Solr where you define your data directories, solr.home etc., DSE does not make use of the node's file system for Solr configuration files. For example, I have this script to setup a Solr core from scratch:
#!/bin/sh
#
# Post files required to define this core
#
CORENAME=Keyspace.CfName # keyspace.column_family
SOLR_HOST=localhost:8983 # host:port
SOLRCORE_URL=http://$SOLR_HOST/solr/resource/${CORENAME}
echo "Posting core configuration files to: ${SOLRCORE_URL}"
curl $SOLRCORE_URL/solrconfig.xml --data-binary @solrconfig.xml -H 'Content-type:text/xml; charset=utf-8'
curl $SOLRCORE_URL/schema.xml --data-binary @schema.xml -H 'Content-type:text/xml; charset=utf-8'
curl $SOLRCORE_URL/synonyms.txt --data-binary @synonyms.txt -H 'Content-type:application/octet-stream; charset=utf-8'
curl $SOLRCORE_URL/stopwords.txt --data-binary @stopwords.txt -H 'Content-type:text/plain; charset=utf-8'
curl $SOLRCORE_URL/protwords.txt --data-binary @protwords.txt -H 'Content-type:text/plain; charset=utf-8'
echo "Created/updated core: ${CORENAME}"
This creates a Solr core named Keyspace.CfName. You can issue Solr queries to DSE using that name, or apply updates etc. Likewise, since you're using Java, as do I, then the Solr Java client library, SolrJ, works as you'd expect. Coming from a separate Solr and Cassandra based setup, I've not yet tried accessing this content, doing searches etc. via CQL. I'm still SolrJ based and it works just fine.
If you use cqlsh or the Cassandra CLI to take a look at what's in that column family, you'll see some metadata columns used to retain the config files uploaded via the API to define the core. Then, each document you add to the core is a row in the column family. So, the Solr configuration and content resides within Cassandra, not the file system. However, the Lucene indicies themselves do reside in each node's file system, but those can be rebuilt if required.
Good luck!
Jeff
