Research On Service Composition Technology In Cloud Computing

In the cloud computing environment,with the diversification and complication of user requirements,the functions provided by a single cloud service can no longer meet the needs of users,and a single cloud service can be combined according to a combination to form a larger granularity of composed services.The service composition is getting more and more attention from experts.However,due to the existence of a large number of services with the same or similar functions in the cloud environment,the selection of a single cloud service and the large amount of time spent in the combination of cloud services make the efficiency of the cloud service composition relatively low.This paper solves the problem of cloud service QoS modeling considerations and the inefficiency of service composition algorithms in the cloud service composition.Firstly,we analyze the factors that affecting the cloud service QoS.The QoS model of a single cloud service is designed by combining the general QoS factor with the load balancing factor.On this basis,we analyzed the relationship between services and services in the cloud computing environment and designed the QoS model of combined services.Further,we derive the mathematical expressions of the cloud service QoS model.Secondly,aiming at the shortcomings of modeling service composition into single-objective optimization problem in the current research,this paper proposes a double optimization goal model with maximizing QoS benefit and minimizing cost,and modeling cloud service composition problem as multi-objective optimization problem.The Pareto solution set is used to solve the multi-objective optimization problem.Finally,aiming at the shortcomings of traditional artificial bee colony algorithm,we proposed an improved artificial bee colony algorithm.We added the discrete effective solution generation strategy based on selection rules and the population selection strategy based on fast non-dominated sorting into traditional artificial bee colony algorithm.The performance of the algorithm is proved by experiments.It is higher than other reference algorithms and is more suitable for cloud computing environments with a large number of services.