SDMS-O: A service deployment management system for optimization in clouds while guaranteeing users’ QoS requirements

In a service oriented cloud, in order to meet users’ functional and non-functional requirements, cloud vendors must manage service deployment effectively. During the deployment process, the cost of deployment and its influence on the cloud are very important issues to consider; however, so far little research has been done to address them. In this paper, we present a Service Deployment Management System for Optimization (SDMS-O) designed with a novel optimization approach for service deployment to improve deployment efficiency and reduce deployment cost while guaranteeing the users’ QoS requirements. In SDMS-O, atom-services as its basic units of service applications are first divided into different service families according to compatibility and installation policy, and a service deployment requirement is expressed as an installation expression sequence. We then present three algorithms to automatically standardize, simplify and optimize this sequence during the service deployment process. Meanwhile, the backtracking technique is applied in each optimization phase of service deployment in order not to violate the users’ QoS constraints. The service deployment result such as safety and usability in each optimization step is also evaluated. A simulated experiment of SDMS-O demonstrates our approach’s effectiveness and efficiency.

► We present a service deployment management architecture.
► We design an approach for simplifying the installation expression sequences.
► The approach can optimize the deployment process.
► Experiments have been done to evaluate the effectiveness of the approach.