Solandra is a real-time distributed search engine built on Apache Solr and Apache Cassandra.
At its core, Solandra is a tight integration of Solr and Cassandra, meaning within a single JVM both Solr and Cassandra are running, and documents are stored and disributed using Cassandra's data model.
Solandra makes managing and dynamically growing Solr simple(r).
For more information please see the wiki
Solandra is relatively stable, most of the common functionality used has been working and stable for users. I personally am no longer developing Solandra much beyond version changes and applying pull requests.
If you are looking for a supported Solr + Cassandra integration beyond what Solandra offers
look at DataStax Enterprise Search
(Full disclosure I am a developer on that team).
I've written up how Solandra and DataStax Enterprise Search differ here.
Java >= 1.6 Cassandra >= 1.1 Solr >= 3.1
The following will guide you through setting up a single node instance of Solandra.
From the Solandra base directory:
mkdir /tmp/cassandra-data
ant
cd solandra-app; bin/solandra
Now that Solandra is running you can run the demo:
cd ../../reuters-demo
./1-download_data.sh
./2-import_data.sh
While data is loading, open the file ./website/index.html in your favorite browser.
To use an existing Cassandra distribution perform the following steps.
Download your Cassandra distribution
Unzip it the directory of your choice
Run the following solandra ant task to deploy the necessary files into the unzipped dir
ant -Dcassandra={unzipped dir} cassandra-dist
You can now start Solr within Cassandra by using $CASSANDRA_HOME/bin/solandra command. Cassandra now takes two optional properties: -Dsolandra.context and -Dsolandra.port for the context path and the Jetty port.
Solandra uses Solr's built in distributed searching mechanism. Most of its limitations are covered here:
http://wiki.apache.org/solr/DistributedSearch#Distributed_Searching_Limitations