Spark is pretty fantastic from our perspective. People just think about it in terms of a faster Hadoop MR but it is so much more. The APIs and integration with external systems are so much easier and more intuitive to use.

It really is Hadoop 2.0.