The Design And Implementation Of The Servicescheduling Subsystem In A Serviced Ynamic Distrubuted Network

As the continuous development of the web service technologies, the service oriented architecture is becoming increasing maturity, and it is becoming more and more concerned in academia and industry. To make the service computing network efficient executed in a dynamic, ever-changing and complex network environment, we optimize it by combining the abstract logical service computing environment with the actual network conditions. But there are still three main problems in the study of service oriented network optimization strategy:the service copy generation problem, that the method of caching service resources static cannot be adapt to complex and changing environment; the service distribution problem, that is because the service net must be optimized through service resources caching and traffic management according to the actual data flow and the logical relationship among services; service resources scheduling problem, that is how to select the optimal service providers and best service copy. In order to solve the problems mentioned above, the author the this paper studied the related themes. And the study of this paper is supported by the National Natural Science Foundation.This paper made an in-depth study of the technologies related to the service dynamic distribution network, especially the basis of the web service, service caching technology and the load balancing technique. These technologies include the composition of the web service, the role and operation of the web service, the classification and distribution method and the consistency of the service caching technology, and the classification, implementation and scheduling algorithms of the load balancing technology. And based on the search of the programs above, this paper has made several important achievements:firstly, it designed and implemented the prototype of the service scheduling subsystem, and the prototype uses a layered architecture, and the core part of the prototype includes three main function modules which are service request scheduling module, service copy generation and distribution module and network performance monitoring module; secondly, it provides a service request scheduling strategy, according to the strategy, it first select the services which meet the functional requirement, and if there are more than one service, then select the optimal service based on the quality of the services, and if there are multiple different copies of the picked service, then we should choose the optimal service copy through the performance of the server on which the service copy is; thirdly, it puts forward a service copy dynamic generation and distribution strategy, by the maintenance of the information that the services are requested and accessed, if the requested numbers exceeds the threshold value set, a copy of the service is generated, and then distribute the service copy to the server which is selected through the server’s performance evaluation method.By using the scheduling subsystem, it is possible to combine the logical environment with the actual network characteristics, distribute the service resources dynamically and make the user’s requests scheduled optimally. Thus it could shorten the waiting time when the user request a service, meet the quality of service requirements of the user, make full use of the entire network resources, as well as enhance the performance of the whole network.