Title :
Prediction-based dynamic load-sharing heuristics
Author :
Goswami, Kumar K. ; Devarakonda, Murthy ; Iyer, Ravishankar K.
Author_Institution :
Center for Reliable & High Performance Comput., Illinois Univ., Urbana, IL, USA
fDate :
6/1/1993 12:00:00 AM
Abstract :
Presents dynamic load-sharing heuristics that use predicted resource requirements of processes to manage workloads in a distributed system. A previously developed statistical pattern-recognition method is employed for resource prediction. While nonprediction-based heuristics depend on a rapidly changing system status, the new heuristics depend on slowly changing program resource usage patterns. Furthermore, prediction-based heuristics can be more effective since they use future requirements rather than just the current system state. Four prediction-based heuristics, two centralized and two distributed, are presented. Using trace driven simulations, they are compared against random scheduling and two effective nonprediction based heuristics. Results show that the prediction-based centralized heuristics achieve up to 30% better response times than the nonprediction centralized heuristic, and that the prediction-based distributed heuristics achieve up to 50% improvements relative to their nonpredictive counterpart
Keywords :
distributed processing; pattern recognition; distributed system; load-sharing; predicted resource requirements; resource prediction; trace driven simulations; Aerodynamics; Delay; Dynamic scheduling; Filters; High performance computing; Length measurement; NASA; Prediction methods; Predictive models; Resource management;
Journal_Title :
Parallel and Distributed Systems, IEEE Transactions on