摘要
首先阐述了Ｗｅｂ服务器性能的定义，然后从Ｗｅｂ服务器性能的几个关键问题入手，从应用负载、服务器软件和服务器硬件３个层面上详细讨论了如何解决这些关键问题．通过对比目前流行的测试工具，选择了合适的应用负载；采用探索式方法优化服务器软件的配置参数，从而减少了测试次数；通过屏蔽各种ＨＴＴＰ请求找出影响性能的瓶颈——对动态请求ＣＡＤ ＧＥＴ的处理．针对该种请求，调整其在动态请求中所占的比例并分析其对不同大小的文件在响应速度上的影响，最后优化相应的代码．将这一方案应用到曙光３０００超级服务器上并使用ＳＰＥＣｗｅｂ９９作为Ｂｅｎｃｈｍａｒｋ测出实验数据，对比改进前的结果，可以明显看到Ｗｅｂ服务器的性能提高了一倍．
The definition of the web server performance is given first in this paper. And then discusses is how to work out the key issues of the web server performance from the three levels of architecture: the application workload, the server software and the server hardware. By comparing the different testing tools, SPECweb99 is chosen as the workload. By using the exploring way, the testing times are greatly decreased. And the bottleneck that affects the system performance is found out by covering the other kinds of HTTP requests: the handling of the dynamic request CAD GET. By tuning its percentage in the dynamic request, its influence on the response time of different files with various length is analyzed and then the corresponding code is optimized. When this scheme is applied to Dawning 3000 super server, the test data shows that 2 times performance improvement can be expected.
出处
《计算机研究与发展》
EI
CSCD
北大核心
2002年第4期415-422,共8页
Journal of Computer Research and Development
基金
国家"八六三"高技术研究发展计划项目基金资助(863-306-ZD01-01)
