Scheduling Job Queue on Hadoop using Hybrid Hadoop Fair Sojourn Protocol

摘要

Background/Objectives: This paper gives an idea how to avoid starvation and minimize response time using hybrid hadoop fair sojourn protocol. Methods/Statistical analysis: The scheduler assigns the resources to the cluster so that job dimension is conditional whereas the job makes improvement towards its conclusion. Scheduling selection help the approach of computer-generated time and cluster funds are focused on jobs according to their arrangement. This ensures that not any of jobs (small or large jobs) undergo starvation. Findings: The Shortest Remaining Processing Time (SRPT)1, ranks the least amount of work to complete, is the one to minimize response time (sojourn time). The Hybrid Hadoop Fair Sojourn Protocol (hybrid HFSP)2 to pause jobs with higher SRPT and allow other waiting jobs in queue based on First Come First Serve (FCFS). Application/Improvements: Relating to existing method the accuracy is developed in suggested method.