In a post, there was a discussion on HDFS vs. KFS. In particular, KFS is integrated with Hadoop's Map Reduce engine.
Out of curiousity, why did KFS not want to create and add a Map Reduce engine in addition to the KFS distributed file system? Why was there the decision to just use Hadoops and integrate it with them?
Is there a reason why Kosmos development didn't want to reinvent the Map/Reduce with KFS and why use KFS with Hadoop Map Reduce?