How to Design a Job Scheduling Algorithm

We discuss the components of a general approach to design algorithms for resource management in parallel processing systems. Starting from the observation that in this area the average influence of research publications on practical implementations is negligible, we address the three main categories constraints, objectives, and evaluation. For each category, we describe common approaches and restrictions and give some general rules that should be followed when presenting a new algorithm for resource management. As an example we present a resource management method for the IaaS model of Cloud Computing that extends the spot instance approach of Amazon. For this example, we first discuss technical, organizational, and usage constraints based on existing concepts and research results. Then we briefly describe resource management objectives from the viewpoint of a provider. After presenting our algorithmic concept, we show that an evaluation with theoretical means can also yield meaningful results in practice.