服务计算环境下一种基于机器学习的负载预测方法研究  

Resarch of a Machine-Learning Based Load Prediction Approach for the Service-oriented Computing Environment

作　　者：王俊[1] 郑笛[1] 吴泉源[1] 官延安[2] 

机构地区：[1]国防科学技术大学计算机学院网络所,长沙410073 [2]中国白城兵器试验中心,白城137001

出　　处：《计算机科学》2007年第9期269-272,共4页Computer Science

基　　金：国家973重点基础研究发展规划(2005cb321804);国家863高技术研究发展计划基金项目(2004AA112020);国家985长江学者和创新团队发展计划(IRT0446)

摘　　要：随着计算机技术的迅速发展,分布式应用的规模迅速增加,越来越多的软件系统开始采用面向服务的体系结构SOA。为了提高SOA的可靠性和可扩展性,一种有效的方式就是提供服务副本,并通过基于中间件的负栽平衡服务在不同的服务副本之间平衡负戢。通过使用中间件,我们可以满足当前的面向服务应用对于性能、可扩展性和可用性的需求。然而,我们必须保证对于负载的计算具有一定的预测性以避免负载峰值的影响。对于复杂的面向服务应用来说,负载峰值意味着系统可能在短时间内具有极高的负载,而在大多数时间内负载较为平稳,从而因为负载取样的延时性导致系统过载而响应时间增加、总体吞吐量也受到影响。因此,为了降低响应时间,以及在负载频繁波动的情况下也能有效地利用服务副本,我们基于中间件为自适应和灵活的负载平衡机制的需求提出并实现了一种基于机器学习的预测机制。With the rapid development of computer technology, the distributed applications scale up increasingly, and more software systems begin to make use of service-oriented architecture SOA. To improve the dependability and scalability of SOA, one effective way is to provide service replicas and balance loads among the replicas via adaptive load balancing service based on the middleware. By using middleware, we can satisfy the urgent demands of performance,scalability and availability in current distributed service-oriented applications. However, we must pay attention to the fact that the computing of the load should be predicative to avoid the affection of the peak load. To the complex service-oriented applications, the peak means the system may suffer extremely high load for a short period which will cause the system to be overload and unstable. The response time will be increased so that the overall throughput will be affected too. Therefore, in order to decrease response time and to utilize the services effectively especially when the workloads fluctuate frequently, we have proposed a new technique based on machine learning for adaptive and flexible load balancing mechanism within the distributed middleware.

关 键 词：面向服务应用 负载预测 机器学习 中间件 

分 类 号：TP393.09[自动化与计算机技术—计算机应用技术]