Hadoop jobs that read from Cassandra seem to run only on master (slaves completely idle)

Question

But when I run the hadoop included wordcount example (dfs version), I see the load getting distributed to all the slaves. What is special about ColumnFamilyInputFormat in Cassandra? Do I need to set any other parameters in hadoop config?

Thanks, Vivek

Answer 1

I would recommend starting with Brisk (http://www.datastax.com/brisk) instead of trying to set up hadoop-on-Cassandra from scratch, since fairly deep knowledge of both systems is required if you take that approach.