Kylin is MOLAP engine on Hadoop, and also possibly on any other distributed computation architecture. With KYLIN-875 <https://issues.apache.org/jira/browse/KYLIN-875>, Kylin supports plugin-able modules of source / engine / storage.
Qianhao has done a great job to demonstrate a Spark build engine this morning. He has developed a Spark engine, together a SparkSQL source module, and build a 10000 records cube in a 3-node Spark cluster. The build completes in 5~ minutes. Will do more serious apple-to-apple comparison between MR and Spark soon.
