Abstract: The open-loop load-balancing algorithm is used to deal with the mixed-pages service requests, which leads to higher load of local servers and reduces the system efficiency in the clustered Web-servers. A closed-loop load distribution strategy based on dynamic optimal adjustment is proposed to solve the problem. It establishes a load model for dealing with mixed-pages containing static pages and dynamic pages according to the characteristic of the service dealing with type of Web request pages and forms an internal dynamic mapping relationship between different service requests and load-balancing to optimize the way of caching and calling static pages. The method based on dynamic prediction of server load rate and equilibrium index adopts an optimal assignment model for task weights with minimum deviation between load rates to calculate optimal task weights for server cluster and achieve load-balancing allocation strategy and algorithm for server cluster dealing with mixed-pages service requests. The simulation experiment results verify the correctness and effectiveness of the load-balancing algorithm proposed in this paper and show that the algorithm can achieve a better load balancing.

Key words: clustered Web-servers, mixed-pages service requests, load-balancing, optimized weigh

摘要： 针对Web集群服务器在处理分配混合页面访问任务时开环负载均衡算法导致局部服务器负载偏高而降低系统效率的问题，建立了一类基于动态调节的闭环负载分配策略。根据服务器处理Web访问页面类型的特点，建立静态页面与动态页面混合处理的负载量模型，形成处理不同服务请求与负载均衡的内在动态映射关系，以此优化静态页面缓存与调用方式，并基于服务器负载率动态预测和均衡指标，采用负载率偏差最小的任务权重最优分配模型，确定服务器集群的最优任务权重，实现了服务器集群处理混合页面访问的负载均衡分配策略和算法。仿真结果验证了负载均衡算法的正确性和有效性，可以达到更好的负载均衡效果。

关键词: Web集群服务器, 混合页面访问, 负载均衡, 优化权重