Cloud adoption by fine-grained resource adaptation: Price determination of diagonally scalable IaaS

Cloud computing is a suitable solution for addressing the uncertainty of resource demand by allowing dynamic resource adjustment. However, most IaaS cloud providers offer their services with a limited granularity at rather slow scaling speeds and flat pricing schemes. Diagonal scaling techniques can offer a more adaptive and fine-grained service with a likewise granular pricing model. Before offering such an adaptive service, cloud providers need a comparison between horizontal and diagonal scaling models to estimate how resource prices can be increased while still staying competitive. In this paper we examine the resource reduction potential of diagonal scaling in comparison to conventional horizontal approaches. Given an empirical load pattern of a web application provider we find a CPU allocation reduction potential of 8.05 % compared to the conventional service. Given a more fine-grained pricing model, we find an additional revenue potential for diagonal scaling of 9.01 % when following a competitor based pricing regime.