Research On SOA-oriented Service Scheduling And Its Application

During the process of Enterprise Information Construction, we usually choose Service Oriented Architecture to integrate the system to avoid the cooperation problems caused by phased transformation or the framework/interface differences between subsystems, it can not only shield the protocol differences between systems, but also make it more convenient to achieve business code reuse. But select the SOA architecture means that the system will communicate and interact by Web Service, in the topology of the whole system, it's likely to come into being a hot service in the central position, which means it's possible to face the dilemma that the request size beyond the service ability. At this point, request scheduling is needed.In different scenarios, request scheduling has different research priorities. When facing the deficiency of service processing capacity, redundant deployment is one solution, it improves the service's processing capacity by expanding the number of service provider. In this scenario, the research priority of request scheduling is how to choose a suitable target in multi service providers to share the burden and achieve load balancing. Another solution is to improve the overall service processing capability by examining the characteristics of request and optimizing the order of requests. The later solution is used in this paper, hoping to propose a scheduling strategy which can improve the success rate of requests and improve the throughput capacity of the service under the premise of no additional deployment and management costs.Firstly, the problem is analyzed, and the research direction of this algorithm is determined by comparing the advantages and disadvantages of the default scheduling strategy and other improved scheduling policies. For SOA integration features, a benefit policy is defined and a set of indicators to benefit is developed under this policy. Based on a historical analysis of the requests, the result effective degree is estimated. The weighted calculation result based on them will be the basis of the request scheduling. By comparing the experimental results, the improved algorithm is verified in the scheduling effect. Finally,application of the algorithm is practiced by applying this algorithm to the practic,al integrated tools.The deficiency of the algorithm is that to a certain extent,the effect of request scheduling is dependent on the difference degree of results computing time when facing different parameters, which means the algorithm is more suitable for complex business services.