Reconfigurable Gang Scheduling Algorithm

摘要

Using a single traditional gang scheduling algorithm cannot provide the best performance for all workloads and parallel architectures. A solution for this problem is an algorithm that is capable of dynamically changing its form (configuration) into a more appropriate one, according to environment variations and user requirements. In this paper, we propose, implement and analyze the performance of a Reconfigurable Gang Scheduling Algorithm (RGSA) using simulation. A RGSA uses combinations of independent features that are often implemented in GSAs such as: packing and re-packing schemes (alternative scheduling etc.), multiprogramming levels etc. Ideally, the algorithm may assume infinite configurations and it reconfigures itself according to entry parameters such as: performance metrics (mean utilization, mean response time of jobs etc.) and workload characteristics (mean execution time of jobs, mean parallelism degree of jobs etc.). Also ideally, a reconfiguration causes the algorithm to output the best configuration for a particular situation considering the system's state at a given moment The main contributions of mis paper are: the definition, proposal, implementation and performance analysis of RGSA.