Interactive computational steering : conservative vs optimistic steering approaches

Abstract

Interactive Computational Steering is the online, interactive allocation and adjustment of system resources and application parameters. Causal consistency is an important feature of interactive steering of distributed computations, as it is often required to maintain the correctness of the computation. However, due to the asynchronous nature of distributed computations, it is difficult to coordinate steering changes across processes to guarantee that the changes are applied consistently at all processes. Two general approaches exist for achieving interactive computational steering: conservative steering and optimistic steering. In this thesis, we present algorithms for the conservative steering approach. We also compare the performance of the conservative steering approach with that of the optimistic steering approach with regard to perturbation and lag.