Web Services Endpoint Monitoring And Modeling Analysis

With the sustainable high speed development and wide application of Internet, distributed computation has achieved great progress. Advancement of distributed computing technology and XML lead to the birth of Web services technology. Web services is loose coupling distributed application that is able to support low-cost, fast, easily assembled, on the basis of loose coupling relationship between services providers and services requesters, with the case that service is computing elements of self-description and neutral platform. Service-oriented computing mode further extends the idea of "software is service". As a service provider, Web services endpoint is also taken as a services requester in the same time. Whether it can normally work or not has an effect on robustness in whole system.The paper concerns about monitoring on load information of Web services endpoint to evaluate present load condition by setting up load prediction model. However, it’s not enough because it has to predict load at next time point on account of monitoring intervals. On the other hand, we can predict when the hard drive capacity is insufficient, which can guarantee normal operation for service endpoint of providing large data storage.To begin with, this paper analyzes Web services characteristic and its most important one is independence between language and platform. Secondly, analyze environment complexity of Web services endpoint. It adopts CPU, memory, network bandwidth utilization, load metrics to evaluate current load situation of the service endpoint as load metrics suite.In a distributed system, in order to ensure the normal runtime of the service endpoint, especially for loading of instantaneous growth, prediction on server load are required. This article uses a server load prediction based on fuzzy control method to establish prediction model of fuzzy control, and fuzzy rules contain trend forecasting strategies, programming with Java implementing the algorithms, and experiments showing good prediction effect. As the Internet grows, more and more information on the Internet, it is quite important for Web service endpoint that provides storing pictures or sound to monitor hard drive storage consumption, whereas the generalization ability of neural network is strong by BP neural network prediction model of hard drive storage consumption.Finally, according to the proposed model and evaluation model, monitoring program of Web service endpoint is realized, making load information from the service endpoint stored in database, and providing information of resources and environment for service endpoint self-management. Its assessment result of monitoring the program costs shows very light impact on the service endpoint.