The purpose of the work in this thesis is to provide the Service Provider a solution which is capable of deploying complex services in a cloud automatically and cost-effectively. The first contribution allows the Service Provider to construct complex services requested by the clients from basic services at his disposal. The construction must be efficient in terms of execution time and operation cost while respecting the client's constraints. We present an analytic model for this problem and propose a heuristic solution which performance is 20-30% better than other approaches. The second contribution solves the problem of deploying the services while considering end-users' demands. To ensure the quality of services provided to end-users, not only one instance but a set of service replicas is deployed in the network. How the service is duplicated and distributed depends on the demands of its end-users that change constantly in quantity as well as distribution, which complicates the problem. Thus the provisioning mechanism has to be capable of adapt to the change in the network, including the change in end-users' demands. Our third contribution is a system using OpenStack which allows Service Provider to deploy complex services in a cloud that spans over different locations (multi-site cloud). Given a client's request, the system automatically calculates the optimal provisioning plan and deploys it while respecting the client's constraints.
