Resources allocation and multi-users task scheduling: a fairness-based approach

Grid and Cloud computing make possible the sharing of computer system resources, such as storage or computation time, among a set of users, according to their requests, thereby creating an illusion of infinite resources. However, as soon as those resources are insufficient to meet users' expectations, conflicts of interest arise. Therefore, unlimited access to limited resources may lead to inefficient usage which penalizes the whole set of users. In such environments, arbitration becomes necessary in order to settle those conflicts and ensure a fair allocation to all users. We present two classes of problems: multi-user resource allocation under uncertainty and multi-user periodic task scheduling. We tackle these problems from the point of view of fairness. (author)