摘要： 针对Hadoop的组合优化问题，提出一种基于混沌粒子群优化算法的配置优化策略。为保证算法在进行局部搜索的同时避免陷入局部最优，对陷入局部最优的标准值和混沌处理的迭代次数的不同组合进行实验，获得使Hadoop性能达到最优的配置。实验结果表明，算法得出的最优配置方案能够有效提高Hadoop的性能，具有较好的稳定性。

关键词: Hadoop平台, 分布式系统, 置方案, 能优化, 子群优化算法, 沌算法

Abstract: Hadoop configuration is a NP-hard combinatorial optimization problem. Aiming at the configuration problem, a configuration optimization strategy is proposed. To ensure that the algorithm not only can search in the local, but also can avoid failing into local optimum, it selects different criteria values and different iteration times to do experiment, so as to search out the values making Hadoop performance optimal. Experimental results show that optimal configuration derived from the algorithm can effectively improve the performance of Hadoop, and has good stability.

Key words: Hadoop platform, distributed system, configuration scheme, performance optimization, Particle Swarm Optimization(PSO) algorithm, chaos algorithm

中图分类号: 

• TP302