Jointly optimizing task granularity and concurrency for in-memory mapreduce frameworks | IEEE Conference Publication | IEEE Xplore