A Neuro Fuzzy Technique for Process Grain Scheduling of Parallel Jobs

Abstract

Problem statement: We present development of neural network based fuzzy inference system for scheduling of parallel Jobs with the help of a real life workload data. The performance evaluation of a parallel system mainly depends on how the processes are co scheduled? Various co scheduling techniques available are First Come First Served, Gang Scheduling, Flexible Co Scheduling and Agile Algorithm Approach: In order to use a wide range of objective functions, we used a rule bases scheduling strategy. The rule system depends on scheduling results of the agile algorithm and classifies all possible scheduling states and assigns an appropriate scheduling strategy based on actual state. The rule bases were developed with the help of a real workload data. Results: With the help of rule base results, scheduling was done again, which is compared with the first come first served, gang scheduling, flexible co scheduling and agile algorithm. The results of scheduling showed the optimized results of agile algorithm with the help of neuro fuzzy optimization technique. Conclusion: The study confirmed that the Neuro Fuzzy Technique can be used as a better optimization tool for optimizing any scheduling algorithm, This optimization tool is used for agile algorithm which is further used for process grain scheduling of parallel jobs.