摘要： 服务器集群在透明计算新模式中的引入，需要多种资源(CPU、内存和I/O等)有效共享。针对传统负载平衡算法考虑资源因素单一、难以满足需求的现状，该文提出一种基于透明计算的负载平衡算法(LBTC)，它充分考虑系统多种资源，并采用负载向量表征和协调。通过不同平均内存需求情况下、不同系统中、不同算法执行复杂任务时的平均减速比较，表明LBTC能以适当增加计算复杂度为代价有效实现资源共享，提高系统性能。

关键词: 网络计算, 服务器集群, 负载平衡

Abstract: For transparence computing, it is in demand to share multi-resources in the server clusters, such as CPU, memory, I/O, etc. To the problem that existing algorithms only consider unitary factor and can’t satisfy the new demand perfectly, a dynamic load balancing algorithm based on transparence computing is presented, shortly named LBTC. It denotes and harmonizes multi-resources with load vector. Results from the mean slowdown compare of different methods in different systems executing jobs with different mean memory demand, show that LBTC with an advisably increased complexity can improve respond time and performance, and meet the demand of transparence computing.

Key words: Network computing, Server cluster, Load balance

中图分类号: 

• TP393