多核Web服务器自适应动态请求调度算法

发布时间：2018-05-11 12:41

本文选题：Web服务器 + 多核处理器　；参考：《小型微型计算机系统》2015年08期

【摘要】：为了提高Web服务器处理用户请求的能力,大多数Web服务器已经采用多核处理器.但传统的动态请求调度算法,例如先到先服务算法,既没有考虑多核处理器的结构特征也没有考虑动态请求服务时间的分布,因此不能充分利用多核处理器的运算能力.为解决此问题,在之前工作的基础上提出一种新的动态请求调度算法.新算法可周期性地预测表征负载状态的动态请求队列的权值,根据预测的权值确定动态请求分配方案,并依据此方案对到达Web服务器的动态请求进行调度.为评估此算法的性能,以此算法为基础进行了仿真实验.实验结果表明,新的算法不仅能解决多核Web服务器中存在的"乒乓"效应等影响性能发挥的问题,而且能在较长时间内保持处理器核心间的负载均衡.
[Abstract]:To improve the ability of Web servers to handle user requests, most Web servers already use multicore processors. However, traditional dynamic request scheduling algorithms, such as first-come first-served algorithms, do not take into account the architecture characteristics of multi-core processors or the distribution of dynamic request service time, so they can not make full use of the computing power of multi-core processors. To solve this problem, a new dynamic request scheduling algorithm is proposed based on previous work. The new algorithm can periodically predict the weights of the dynamic request queue representing the load state, determine the dynamic request allocation scheme based on the predicted weights, and schedule the dynamic requests arriving at the Web server according to this scheme. In order to evaluate the performance of the algorithm, simulation experiments are carried out on the basis of the algorithm. Experimental results show that the new algorithm can not only solve the problem of "ping-pong" effect in multi-core Web servers, but also maintain load balance between processor cores for a long time.
【作者单位】：北京化工大学信息科学与技术学院信息中心;
【基金】：中央高校基本科研业务费项目(YZ1319)资助
【分类号】：TP393.09

【参考文献】