At Hack Reactor, we are setting up a computationally heavy MapReduce job across a 1,000 machine cluster. It is an algorithmic tree traversal expected to run over several hours, comprising over 10 quintillion computations, and occupying at least a few petabytes of storage across HDFS.
We plan to run the job using a Java implementation of our functions. The original, however, is in JavaScript, and I noticed that JavaScript UDFs are still considered experimental for want of additional testing. If our project could be of any use in testing edge cases or proving out the JavaScript UDF functionality in Pig, we would be more than happy to help. Cheers, RP