A Comprehensive Study of Multi-Scheduling Schemes Performance

다국어 초록 (Multilingual Abstract)

Today distributed server systems have been widely used in many areas because they enhance the computing power while being cost-effective and more efficient. Meanwhile, efficient multi-scheduling schemes are employed to optimize the task assignment process. This paper closely explored the performance of multi-scheduling schemes through computer simulation. The research was started regarding the simulation of a novel scheduling policy (Task Assignment by Guessing Size) associated with other two previous task assignment policies (Random and JSQ). The multi-scheduling schemes involve two types: Random-TAGS scheme and JSQ-TAGS scheme. To facilitate the performance, computer simulation is applied to perform the statistical measurements. The findings were, indeed, very interesting, showing that the multi-scheduling scheme obtains better performance than single scheduling strategy scheme under heavy-tail distributed computing environment. Furthermore, JSQ-TAGS scheme is more efficient and stable in contrast to Random-TAGS scheme. The paper finally concludes by summarizing the findings from the simulation and suggesting a wider study be undertaking, in order to explore the performance of multi-scheduling schemes in more depth.

목차 (Table of Contents)

• Abstract
• 1. Introduction
• 2. Scheduling Strategies
• 2.1. Join the Shortest Queue policy
• 2.2. Random Assignment Policy
• 2.3. Task Assignment by Guessing Size
• 3. Performance of Single Scheduling Schemes
• 3.1. Performance Based on Exponential Distribution
• 3.2. Performance Based on Pareto Distribution
• 4. Design of Multi-Scheduling Schemes
• 5. Measurements and Analysis
• 6. Conclusion
• Acknowledgements
• References