Study how to develop support for specific application types and application areas in schedulers for large- scale systems such as grids and clouds.
Different application types, both old (e.g., workflows) and new (e.g., mapreduce and data- mining at unprecedented scales), may want to take advantage of the capabilities of evolving large-scale computing infrastructures, such as (inter-operating, federated) grids and clouds with their heterogeneous hardware (processor types) and high-speed networks. No single scheduler can incorporate support for all imaginable types of applications for all types of computing infrastructures. Therefore there is a need to design and build application-support modules in schedulers of large-scale distributed systems.
Develop support in the KOALA scheduler for specific application types and for applications in specific application areas (e.g., life sciences) that are deployed in grids and clouds, and, in particular, that are studied in COMMIT. We will extend the current KOALA scheduler with application-support modules for these application types and areas in such a way that users do not have to deal with details of the underlying systems but can concentrate on application design, and that applications can run (energy-)efficiently and cost-aware.