Kind of a long blog, but interesting as always from Dr. Mattson. Check out point #3 towards the end of the entry. Dr. Mattson states that "Domain specific languages may be our best hope.". I think he is onto something there. Look at Java7 and the new concurrent library features that are being added: Fork-join and parallel arrays. These are both very useful constructs for helping to parallelize sections of your application code. But, being general in use, they are meant for in-memory only data structures. Very useful, but limited for large data applications where billions of events must be processed.

DataRush is built with data scalability in mind. By that I mean you can run a DataRush application with thousands, millions or billions of rows with little to no adjustment and no re-coding. DataRush is domain specific; the domain being very large data analytics. Have billions of rows of data to process and analyze: DataRush can help!

DataRush does not implement a new language, but is built on the existing concepts of dataflow architecture. Our goal is to provide a Java based set of libraries that allow programmers to build highly performant, highly scalable data analytic applications easily. All this using a language you are already familiar with and already enjoy great IDE support.