One of the more interesting analytics stacks being built on top of Hadoop is coming out of UC Berkeley.
Particularly interesting is SparkR, which they just blogged about here:
We don't want to get ahead of ourselves; after all, we're just starting to get some page view data into HDFS, but it's important to understand that one of the reasons we like Hadoop is the ecosystem of open source tools built around it.
-Toby