Virtual resource auction based on Bayesian incentive strategy in large-scale clouds

Abstract

In cloud platforms, resource pricing service plays a key role to regulate the behaviours of both resource providers and consumers. However, the increasing diversity of user quality-of-service (QoS) requirements makes existing pricing models difficult to be implemented in an efficient manner. In this paper, we design an auction model which is not only useful for cloud clients but also can significantly increase the resource revenue for providers. To support QoS-aware resource pricing, we normalise QoS parameters-based user's scores and use the Bayesian incentive strategy to regulate resource auctions. The key advantage of this auction model is that it supports multi-attributes auction and budget-balancing among bidders. Extensive experiments are conducted in a campus-based cloud, and the results are compared with other existing pricing models. The results indicate that the proposed auction model can significantly improve the resource revenue of cloud providers as well as maintain desirable QoS level for cloud clients.