Predictive job scheduling in a connection limited system using parallel genetic algorithm

Job scheduling is the key feature of any computing environment and the efficiency of computing depends largely on the scheduling technique used. Intelligence is the key factor which is lacking in the job scheduling techniques of today. Genetic algorithms are powerful search techniques based on the mechanisms of natural selection and natural genetics. Multiple jobs are handled by the scheduler and the resource the job needs are in remote locations. Here we assume that the resource a job needs are in a location and not split over nodes and each node that has a resource runs a fixed number of jobs. The existing algorithms used are non predictive and employs greedy based algorithms or a variant of it. The efficiency of the job scheduling process would increase if previous experience and the genetic algorithms are used. In this paper, A new technique is proposed as a model of the scheduling algorithm where the scheduler can learn from previous experiences and an effective job scheduling is achieved as time progresses.