Decentralized Resource Coordination across Service Workflows in a Cloud Environment

Abstract

As Cloud computing becomes one of the mainstream distributed computing paradigms, cloud users are generating more intricate business processes and scientific jobs which involve the use of service workflows. Efficient management of those workflows' resources can help save cost and improve QoS. In this paper, we investigate the problem of coordinating resources beyond the boundary of a single workflow. An agent-based decentralized algorithm is proposed to match resource supply and demand from different workflows. The proposed algorithm uses multiple phases of information exchange and local adjustments to produce conflict free resource reallocation decisions. Experiments are conducted using simulation programs and synthetic data. Comparisons with a centralized method demonstrate the effectiveness of the proposed algorithm.