Model-Based Automation of Service Deployment in a Constrained Environment

Service physical topology design in large data centers is a function of many cross-cutting concerns such as service connectivity, performance, and security requirements, as well as data center resources, policies and best practices. These interdependencies represent a significant source of complexity, cost, and risk in data center management. This paper proposes a novel model-based approach to service physical topology design that reduces design complexity and allows for a separation of concerns. Service requirements are represented in terms of highlevel models, and provisioning concerns as model transformation rules, resource selectors, constraints and objective functions. We use a refinement engine that automatically searches the space of service model transformations to produce a set of possible service physical deployment topologies. We discuss our prototype implementation of the service model refinement engine design and we report the results of experiments where we used our prototype to generate detailed physical deployment topology solutions for a variety of service models describing a rich set of requirements.