Briefing - Dynamic Workers for Scheduling

7

Click here to load reader

Transcript of Briefing - Dynamic Workers for Scheduling

Page 1: Briefing - Dynamic Workers for Scheduling

Dynamic Workers for SchedulingAuthors: 唐子晴、邱柏誌Advisor: Chien-Wen Wu

Page 2: Briefing - Dynamic Workers for Scheduling

Introduction

•Scheduling is essential for backend services

•Adoption of “Energy Saving System”•The paper introduces the dynamic

workers to finish the queued tasks by closing the idle workers depending on the assigned tasks.

Page 3: Briefing - Dynamic Workers for Scheduling

Literatures

•Simple Linux Utility for Resource Management, Power Saving Guide▫https://

computing.llnl.gov/linux/slurm/power_save.html

•Scheduling Algorithms by Professor Dr. Peter Brucker

•A Particle Swarm Optimization (PSO)-based Heuristic for Scheduling Workflow Applications in Cloud Computing Environments

Page 4: Briefing - Dynamic Workers for Scheduling

Models

•Combination of “Energy Saving System” and tasks assignment algorithm to identify the closing idle workers.

•Sorting algorithm for importance of jobs (and each job complexity) for deciding down time of the idle workers.

Page 5: Briefing - Dynamic Workers for Scheduling

Algorithms

•Particle Swarm Optimization(PSO)▫Solving task assignment problem

•Bucket sorting

Page 6: Briefing - Dynamic Workers for Scheduling

Conclusion

•We expect to achieve a better energy saving solution for server nodes (workers)

•A properly assigned computing power for load-balancing

Page 7: Briefing - Dynamic Workers for Scheduling

Thanks for Listening!