Multi objective Virtual Machine Resource allocation by Dynamic Load scheduling techniques

Abstract

With ease of internet facilities number of jobs were handle by servers where jobs are newlineof dynamic in nature. Hence dynamic load balancing of the work need to be newlinerequired. Researcher has proposed number of techniques for this issue but jobs newlinepattern identification for reducing the balancing time is highly required. newlineThe objective of this work was to take heterogeneous jobs as input and find the job newlinesequence which can reduce time. Dynamic load balancing was done by introducing a newlinegenetic algorithm which can adopt the dynamic situations. newlineIn this work grouping of various sort of job was done by proposed Teacher Learning newlineParticle Swarm Optimization and Random Forest Dynamic Load Balancing, job newlinesequencing models. In order to make a general model which work on various newlineavailable data Indices a Teacher Learning Particle Swarm Optimization (TLPSO) newlinegenetic algorithm was used where data is classified without finding all possible newlinecombination. Here dynamic load was easily scheduled by this TLPSO where hybrid newlinecombination has increase the work efficiency. This work has developed a model newlinewhich can adopt dynamic situation by using genetic algorithm and learn the patterns newlineof job requirement as well. newlineMost of time jobs have the same set of resource requirements for learning random newlineforest learning technique was used which is a combination of job positional decision newlinetrees, where input is set of resource requirement as per job and output is sequence newlineof jobs to be executed. While dynamic jobs sequences were managed by Butterfly newlineAlgorithm Particle Swarm Optimization. So combination of both these techniques newlinewas termed as RFDLB (Random Forest Dynamic Load Balancing). newlineIt was also shown that use of genetic algorithm alone has not performed well for newlinegenerating the sequence. Random forest has collection of decision trees which take newlinesuperior decision as per training sequence and job machine requirements. newlinePerformance of load balancing algorithms was done by improving PSO algorithm in newlinecrossover operation of teacher learning