Research Article
On Cost Modeling for Hosted Enterprise Applications
@INPROCEEDINGS{10.1007/978-3-642-12636-9_18, author={Hui Li and Daniel Scheibli}, title={On Cost Modeling for Hosted Enterprise Applications}, proceedings={Cloud Computing. First International Conference, CloudComp 2009 Munich, Germany, October 19--21, 2009 Revised Selected Papers}, proceedings_a={CLOUDCOMP}, year={2012}, month={5}, keywords={}, doi={10.1007/978-3-642-12636-9_18} }- Hui Li
Daniel Scheibli
Year: 2012
On Cost Modeling for Hosted Enterprise Applications
CLOUDCOMP
Springer
DOI: 10.1007/978-3-642-12636-9_18
Abstract
In enterprises nowadays typical business-critical processes rely on OLTP (online transaction processing) type of applications. Offering such applications as hosted solutions in Clouds rises many technical and non-technical challenges, among which TCO (Total Cost of Ownership) is one of the main considerations for most on-demand service/Cloud providers. In order to reduce TCO, a first step would be to analyze and study its cost components in depth. In this paper we adopt a quantitative approach and model two tangible cost factors, namely, server hardware and server power consumption. For server hardware, on one hand, a pricing model for CPU is proposed as a function of per-core performance and the number of cores, which also manifests the current multi-/many-core trend. Server power consumption, on the other hand, is modeled as a function of CPU utilization (as a main indication of system activity). By using published results from both vendor-specific and industry-standard benchmarks such as TPC-C, we show that a family of is successfully applied in deriving a wide range of cost models. Such analytic cost models, in turn, prove to be useful for the Cloud providers to specify the Service Level Agreements (SLAs) and optimize their service/infrastructure landscapes.
