Qos aware resource scheduling and evaluation of heuristics for processing big data on cloud

Abstract

Results and Discussion: In this research, we have simulated First Fit and Best Fit heuristics with various SLAs like Hard SLA, Best Effort SLA, and Soft SLA. Jobs with required QoS parameters like Reliability, Execution Time and Priority are assigned to the best-matched resource according to heuristic and SLA. Quality of resource is determined by parameters like Reliability, Job Completion Time and the Cost of the resource. Performances of both the heuristic approaches are evaluated with performance parameters like Average Resource Utilization (ARU), Success Rate of Jobs (SR) and Total Completion Time (TCT). newline newlineConclusions: Best Fit heuristic will perform slightly better than the First Fit for ARU and SR. But First Fit will perform better for TCT in all simulation environments due to the early allocation of resources. The highest number of jobs will get scheduled in Soft SLA due to the weakest restrain and Lowest in the case of Hard SLA due to the hardest restrain which will cause an increase in ARU, SR while the decrease in TCT. So, Soft SLA will perform better for both heuristics and in all simulation environments but Job will have to compromise the most in Soft SLA while Hard SLA will not perform that much good compare to Soft SLA as far as the number of jobs which will get scheduled is concerned. But in Hard SLA jobs will have to do the lowest compromise for their QoS requirements. Performance of Best Effort SLA is in between Hard SLA and Soft SLA for both the heuristics. This research work is useful for various organizations that provide various Cloud services to users who seek different levels of QoS for various applications. newline newline