中文摘要：

首先根据排队论理论给出了一个理想的负载均衡模型。针对该模型的系统开销问题，提出了一种基于节点分组的异构集群负载均衡算法。实验结果表明，相比于轮转法和加权轮转法这两种普遍采用的负载均衡算法，该算法可以提供更加稳定的请求响应时间，并在负载波动较大时明显提高集群的吞吐率。

英文摘要：

This paper first proposed an ideal load balancing model based on queueing theory, then described a nodes partition algorithm of heterogeneous clusters in order to solve the performance problem. Experimental results show that compared with the round robin and weighted round robin policy, the algorithm can provide better response time, and improve throughput greatly under highly different workloads.