Evaluator services for optimised service placement in distributed heterogeneous cloud infrastructures

Optimal placement of demanding real-time interactive applications in a distributed heterogeneous cloud very quickly results in a complex tradeoff between the application constraints and resource capabilities. This requires very detailed information of the various requirements and capabilities of the applications and available resources. In this paper, we present a mathematical model for the service optimization problem and study the concept of evaluator services as a flexible and efficient solution for this complex problem. An evaluator service is a service probe that is deployed in particular runtime environments to assess the feasibility and cost-effectiveness of deploying a specific application in such environment. We discuss how this concept can be incorporated in a general framework such as the FUSION architecture and discuss the key benefits and tradeoffs for doing evaluator-based optimal service placement in widely distributed heterogeneous cloud environments.