Resource Allocation in Computational Grids - A Market Engineering Approach

Abstract

The Computational Grid is a promising technology for providing access to distributed high-end computational capabilities. However, deciding which jobs are allocated to which resources is one the key problems in Computational Grids. In recent times, researchers have increasingly suggested to employ market mechanisms for scheduling and allocating Grid resources. This paper outlines the design of a market for allocating and scheduling resources in Computational Grids. The design of the market is based on the Market Engineering process, which provides methods and design principles for the development of market institutions. First, the characteristics and the requirements of the market participants as well as the Grid resources are elicited. Furthermore, a combinatorial clearing formulation is introduced as a primal mixed integer problem, which supports bids on bundles, quality, and time attributes. Followed by a performance simulation, the runtime of the clearing model is simulated.