Quite interesting article on Gigaom.com which says that Cloudera is developing a system called Oryx. The system is aiming to be a better Mahout.
The things that are supposed to differentiate it from Mahout are mainly:
- It will not only provide means for exploratory analysis of data but provide tools for deploying production services containing models produced by machine learning algorithms as well.
- It will not only be based on MapReduce, but will use Apache Spark as well (Spark is a more and more popular Hadoop-like technology).