We are in the process of evaluating Cassandra for our needs and we expect to use parallelism for analytics on collected data. The following BLOG mentions that
Cassandra's Map-Reduce support does not support multiple column families http://allthingshadoop.com/2010/04/24/running-hadoop-mapreduce-with-cassandra-nosql/

Has the same proven to be a challenge or blockers for those using the Map-Reduce functionality? Please share your experience. Also are there plans to support multiple column family map-reduce functionality?