Scheduling for Large Scale Distributed Computing Systems: Approaches and Performance Evaluation Issues

Although our everyday life and society now depends heavily oncommunication infrastructures and computation infrastructures,scientists and engineers have always been among the main consumers ofcomputing power. This document provides a coherent overview of theresearch I have conducted in the last 15 years and which targets themanagement and performance evaluation of large scale distributedcomputing infrastructures such as clusters, grids, desktop grids,volunteer computing platforms, ... when used for scientific computing.In the first part of this document, I present how I have addressedscheduling problems arising on distributed platforms (like computinggrids) with a particular emphasis on heterogeneity and multi-userissues, hence in connection with game theory. Most of these problemsare relaxed from a classical combinatorial optimization formulationinto a continuous form, which allows to easily account for keyplatform characteristics such as heterogeneity or complex topologywhile providing efficient practical and distributed solutions.The second part presents my main contributions to the SimGrid project,which is a simulation toolkit for building simulators of distributedapplications (originally designed for scheduling algorithm evaluationpurposes). It comprises a unified presentation of how the questions ofvalidation and scalability have been addressed in SimGrid as well asthoughts on specific challenges related to methodological aspects andto the application of SimGrid to the HPC context