The Analysis And Realization Of The Strategy Of Clouds Services Selection Based On CDMI

Many manufacturers begin to put forward the concept of cloud storage which is making use of a large number of storage equipments, cloud storage combines virtualization technology, data storage and so on. Among the cloud storages, cloud storage provides various quality of services, how to choose the most suitable for the user's own needs from multiple cloud storage has become more urgent in current research.By comparing Apache LibCloud and OpenStack Cloud Data Management Interface(CDMI), CDMI is better than LibCloud in developing and maintenance. The platform provides the capability detection and service selection function. Namely, capability detection is to detect several capabilities of clouds periodically, including speed, throughput, costs and so on and service selection is to decouple the users and cloud storages. System can get the quality of service of these clouds based on the data. Service selection which use the exponential weight is to select a most suitable cloud storage for the user when user sends a require. By realizing the capability detection function, service selection function and the basic concept of CDMI, the CDMI-based cloud combination system is built.In functional test, capability detection and service selection works well and can give desired results. By comparing two methods' building virtual space time, smoothed and non-smoothed speed curve, and the time and the accuracy among the three decision algorithm in the performance test, from the results, the convergence time of the system decrease 30% or more compare to the origin-based algorithm, and smoothed curve has less jitter than not smoothed curve, rate prediction results range decreased average 50%, fully meet the needs of users.