Towards an Efficient Decision Policy for Cloud Service Providers

Cloud service providers may face the problem of how to price infrastructure services and how this pricing may impact the resource utilization. One aspect of this problem is how Cloud service providers would decide to accept or reject requests for services when the resources for offering these services become scarce. A decision support policy called Customized Bid-Price Policy (CBPP) is proposed in this paper to decide efficiently, when a large number of services or complex services can be offered over a finite time horizon. This heuristic outperforms well-known policies, if bid prices cannot be updated frequently during incoming requests and an automated update of bid prices is required to achieve more accurate decisions. Since CBPP approximates the revenue offline before the requests occur, it has a low runtime compared to other approaches during the online phase. The performance is examined via simulation and the pre-eminence of CBPP is statistically proven.