As use of clusters/compute grids becomes more diversified, the industry is witnessing a convergence of high-throughput computing (HTC) with high-performance computing (HPC). While HPC workloads are compute/data intensive and can take from hours to months to complete on multiple nodes, HTC jobs tend to have short runtimes, often in the millisecond to minutes range on a single node or core. The Nearly all HTC jobs can be classified as “embarrassingly parallel,” meaning the workload can be divided up into multiple, autonomous serial pieces, each of which can be independently executed. Other workloads include regression tests, job arrays, or other tasks which create large queues. SOA workloads have many similar needs