Cloud Service Model Of Service Oriented Uncertainty Demand

With the rapid development of cloud computing and Internet technology, software for users to use by people of all ages in the form of service. In the cloud computing environment integrates a massive variety of services, users can purchase the required service according to the actual demand. Massive service has many of the same features but non functional attributes Qo S values of different services and cloud computing for multiple users, each user’s demand function is diverse, requirement description is also a variety of non-standard, there is ambiguity. The uncertain demand makes it difficult to accurately provide users to meet the needs of the service, has become an urgent need to solve the problem.In view of the above problems, this paper carries on the research from the following several points:In this paper, a cloud model based on Pi algorithm and Spark ant colony algorithm is proposed, which is based on the requirement of users, the service flow and service generation. The cloud model is designed as the three layer, which is the requirement description layer, the service flow layer and the service layer.The first layer is the requirement description layer: in the first layer, the service ontology description language OWL-S is used to formally describe the user’s requirements,and get the requirement documents that can be recognized by the computer without two meanings. As users continue to increase, the demand will become large and complex documents, the use of cloud computing data processing technology to deal with MapReduce, reduce the need for redundancy and conflict. Define the Pi calculus and OWL-S language mapping rules, the OWL-S description of the requirements of the document and the Pi calculus mapping, formal dynamic description of the process of correctness, the output result driven service process layer.The second layer is the service process layer: using the memory computing framework Spark to improve the ant colony algorithm, improve its efficiency, and get the parallel ant colony optimization algorithm, which is better in the use of cloud computing distributed environment. Demand layer uses OWL-S to describe the demand to accordingto the functional requirements of the automatic discovery to find many of the same features but non functional attributes Qo S values of different services, in these services using improved spark parallel ant colony algorithm based on quality of service(QoS) value to identify the optimal atomic services; using PI calculus of OWL-S to describe business needs were formal description, service process. The output of the driving service layer.Third layer service layer: according to the service flow layer to get the best service and service flow, using Pi algorithm to realize the communication between the services, so as to generate new services to meet the needs of users.In the Hadoop2 platform, using yarn data management framework, integrated spark of large distributed data processing technology, through customer relationship management service as an example to verify the service oriented uncertain demand of cloud model is feasible and effective, to achieve in the complex cloud environment fast accurately for each user to provide services to meet the needs of.