摘要： 分布式系统的多用户环境、拓扑结构、并发操作等因素影响其运行性能，且在测试中很难找出制约系统性能的瓶颈资源。为解决该问题，提出一种自适应遗传算法。逐步增加负载压力，根据响应时间定位系统瓶颈资源所在部位，利用采集到的服务器关键性能指标值组成染色体，基于各指标变化对响应时间的影响程度构建系统适应度函数，进行遗传变异操作。实验结果表明，该算法能准确找出分布式系统的性能瓶颈资源。

关键词: 分布式系统, 性能瓶颈, 自适应遗传算法, 关键性能指标, 适应度函数

Abstract: Distributed software’s run-time performance is vary largely by such factor as multi-user environment, the topology structure and parallel operations besides it is difficult to find out the system’s deficit. Aiming at the problem, this paper presents an adaptive Genetic Algorithm(GA). It adopts the method that to increase the load gradually, according to the response time check which parts in the system occurs performance problem, then forms the chromosome by using the collected key performance index value. The fitness function is created based on the fact that along with the vary of multiple indexes which affects the response time differently, and do inherit and mutation operation to find out the factor. Experimental result shows that this algorithm can accurately find out the system performance bottleneck resources.

Key words: distributed system, performance bottleneck, adaptive Genetic Algorithm(GA), key performance index, fitness function

中图分类号: 

• TP39