Two risk-aware resource brokering strategies in grid computing: Broker-driven vs. user-driven methods

Abstract

Grid computing evolves toward an open computing environment, which is characterized by highly diversified resource providers and systems. As the control of each computing resource becomes difficult, the security of users' job is often threatened by various risks occurred at individual resources in the network. This paper proposes two risk-aware resource brokering strategies: self-insurance and risk-performance preference specification. The former is a broker-driven method and the latter a user-driven method. Two mechanisms are analyzed through simulations. The simulation results show that both methods are effective for increasing the market size and reducing risks, but the user-driven technique is more cost-efficient. © 2010 Springer-Verlag Berlin Heidelberg.