Resource unaware load distribution strategies for processing divisible loads in networked computing environments

Abstract

In this thesis, we first consider processing divisible loads in resource unaware linear and multi-level tree networks. A probing technique is used to detect the link and processor speeds, which are then used by the scheduler to generate a feasible schedule. We then argue the usefulness of probing technique in arbitrary networks and/or when multiple sources exist. An alternative reporting based technique is suggested. We also study and compare the performance of different spanning trees in scheduling divisible load(s) in arbitrary networks. Finally, scheduling multi-source divisible loads on arbitrary networks is addressed. Starting from resource aware case, we propose efficient strategies to schedule multi-source loads in two different cases - when no new loads arrive at the system and when new loads may arrive as time progresses. We also demonstrate that by using a reporting based scheme, our strategies can be easily adapted to the resource unaware case.