广域Web集群的随机高级Petri网模型及性能分析

广域Web服务器集群是由多个局域Web服务器集群在Intemet上广域分布而形成的一种服务器体系结构,是目前解决Web系统超载和提供高性能服务器的一种有效手段.基于随机高级Petri网的性能建模与分析技术,提出了一种广域Web服务器集群的随机高级Petri网模型;对模型进行了精化设计以简化模型求解的复杂性;研究了广域Web服务器集群的多级负载均衡综合方案,包括广域集群级的集群站点选择算法、局域集群级的请求分配算法和服务器级的请求调度算法,这些算法可以一致地综合在随机高级Petri网模型中;并且利用随机高级Petri网的性能分析技术对负载均衡综合策略进行了性能分析与评价.

Modeling and Performance Analysis of Globally Distributed Web-server Clusters using Stochastic Highlevel Petri Nets

A globally distributed Web-server cluster consists of multiple local Web-server clusters geographically distributed on the Internet. It is presently a promising solution to the problem of overloading of Web-server systems. Based on the performance modeling and analysis technique of stochastic high-level Petri nets (SHLPN), this paper first propose a SHLPN model for the globally distributed Web-server cluster; the SHLPN model is further refined to reduce the complexity of the model; the multilevel load-balancing policy is investigated, which is integrated by the local-cluster selection algorithm at the global cluster level, the request dispatching algorithm at the local cluster level, and the request scheduling algorithm at the Web server level; the performance of the integrated load-balancing policy is analyzed and evaluated using the performance analysis technique of the SHLPN.