Resource Management for Application Frameworks in Clouds - RMAFCloud

Over the last decade, a number of programming models and application frameworks such as MapReduce, Pregel, Dryad and Big-Data analytics have emerged as popular paradigms in large clusters, datacenters, and clouds. Rather than scheduling single applications, the resource managers and schedulers in such systems should schedule the frameworks, taking such constraints as fairness and data availability into account. In this project, the software for incorporating this functionality into the KOALA multi-cluster scheduler will be designed, implemented, and deployed in the Dutch DAS4 multi-cluster system and in selected clouds, and experiments will be executed with it to assess its performance. One of the main objectives of this project is to design a generic framework that allows a wide range of current and future frameworks to be scheduled dynamically in cluster and datacenter environments.