Hadoop中任务调度算法的改进

摘要

针对Hadoop0．20．0中任务调度算法存在的不足，提出一种基于改进遗传算法（IGA）的任务调度算法。IGA算法对初始化种群、交叉和变异操作进行了一些改进，并引入了最优保留策略和加速进化策略。通过仿真实验将此算法与Hadoop现有算法进行比较，实验结果表明，此算法优于Hadoop现有算法，是一种有效的任务调度算法。%A task scheduling algorithm based on improved genetic algorithm (IGA) was brought up for the deficiences of the task scheduling algorithms in Hadoop0.20.0. IGA algorithm improved the original colony is initialized and crossover and mutation operation, and introduces retention optimal strategy and evolution acceleration strategy. There is a contrast between IGA and existing algorithm in Hadoop through simulation experiment, and the result is: the IGA is better, it is an effective task scheduling algorithm.