Storm [1] and Spark [2] are two other distributed stream processing libraries in this family. Lots of good work being done in this area; batch-processing a la Hadoop is simply unnecessary for a lot of incredibly useful algorithms.
As is Streams [1], which I work on. There are many research papers on the project which use the research name System S to differentiate active research from the product.