Application Skeletons: Encapsulating MTC Application Task Computation and I/O

Computer scientists who work on tools and systems meant to support or enable a variety of distributed computing appli-cations want to prove that the systems they design actually help those applications. However, doing this by using the actual applications can be difficult due to policy or techni-cal issues when accessing and building the application and necessary data sets. These issues led us to the idea of an Application Skeleton – a simple yet powerful tool to build synthetic applications that represent real applications, with runtime, I/O, and intertask communication close to those of the real applications. This allows computer scientists to fo-cus on the system they are building; they can work with the simpler Skeleton applications and be sure that their work will also be applicable to the real applications. Skeletons currently can create easy-to-access, easy-to-build, and easy-to-run bag-of-task, map-reduce, and multi-stage workflow applications. In this initial work, we show that a Skeleton version of the Montage application has a runtime difference of 2.6 % in total on 64 processors on a BG/P supercomputer. And six of eight stages have an error within 5%. 1