文档介绍：燕山大学工学硕士学位论文IIAbstractWith the increasing demand for puting speed, high systemreliance and high performance, apparently, there must be puterstructures take the place of the traditional Von Neumann structure. The parallel and distributed system is a most excellent choice. Owing to its features, such as depressing the bottleneck of processing, providing better ratio of performance to price, studies and applications on distributed system have e more in-depth and extensive. This paper focuses on the research of load balancing in puter , this paper finds out the reason to extra cost to load balancing andthe difficulties of load balancing, which is based on the study of existing load balancing algorithms and models applied in then, brings out information center load balancing policy, which synthesisthetraditional load balancing policies, and is more adaptive in , this paper gives out several implementation models, such as the chain model,thereticulation model andthe chain-reticulation model. Eachmodel is given its corresponding algorithm and evaluation, which draws the conclusion that the chain-reticulation model has a more excellent performance in dynamic load , this paper gives out information center load balancing model by Java RMI paper takes efforts to study the load balancing problem that is an optimizing technology in puter puting; Load Balancing; Dynamic Load Balancing; Threshold; 选题背景随着网络技术的迅猛发展人们越来越多地使用网络网络访问量也不断激增网络的各个核心部分随着业务量的提高和数据流量的快速增长对其处理能力和计算强度的要求不断增大这使得单一设备根本无法承担过大的工作量[1]如果采取硬件设备升级方案将造成现有的资源浪费同时单个硬件的处理能力有限硬件升级方案不能根本解决这类问题由于关键业务的网络化和网络访问的突发性特点需要寻求强有力的方式来获取高可用性和可伸缩性的网络服务器而对于一个分布式计算机系统由于任务到达的随机性以及各处理结点处理能力上的差异当系统运行一段时间后某些节点分配的任务还很多而另一些节点却是空闲的一方面使超载节点上的任务尽可能快地完成是当务之急另一方面使某些节点空闲是一种浪费实际应用中需要避免这种空闲与忙碌并存地情况从而有效地提高系统地资源利用率减少任务地平均响应时间正是在这样的情况下负载均衡机制应运而生负载均衡建立在现有网络体系结构之上它提供一种有效并且透明的方法来扩展网络设备的带宽增加吞吐量加强数据处理能力提高网络的灵活性和可用性负载均衡主要解决两方面的问题第一个是将大量的并发访