摘要
传统的集群服务器依靠前端分发器基于负载均衡来分发客户请求,而ASAS集群采用了基于后端服务器负载分析后的主动调度策略。控制ASAS集群后端服务器的CPU利用客户请求的响应时间,或者连接容量,却很难建立一个精确的数学模型,因此传统的控制器很难控制。这里设计了一个模糊控制算法,控制算法的输入是CPU利用率和客户请求的响应时间,输出是连接数,试验表明这种模糊控制策略取得了很好的性能表现。
In traditional cluster server, the front- dispatcher distributes clients＇ requests by load balancing, but ASAS（Automatic Self- Al： locating Server） duster depends on its back- server＇ s Automatic Self- Allocating based on load status. Because of the difficulty to build a precise mathematical model in the back - server of ASAS cluster to control CPU usage and response time of clients＇ request or link capacity, traditional controller is hard to manage, so we designed a fuzzy control algorithms. Its inputs are CPU usage and response time of clients＇ request, its output is link number, tests show that this fuzzy control method achieves good performance.
出处
《计算机技术与发展》
2006年第12期39-42,共4页
Computer Technology and Development
关键词
集群
主动自调度服务器
模糊控制
cluster
automatic self - allocating server
fuzzy control
