Browse by author
Lookup NU author(s): Dr Christopher Smith, Professor Aad van Moorsel
The emergence of the cloud computing paradigm promises flexibility and adaptability through on-demand provisioning of compute resources. As the utilization of cloud resources extends beyond a single provider, for reasons including fault-tolerance and performance, the issue of effectively managing such resources comes to the fore. Different providers expose different interfaces to their compute resources utilizing varied architectures and implementation technologies. This heterogeneity poses a significant system management problem, and can limit the extent to which the benefits of cross-cloud resource utilization can be realized. We address this problem through the definition of an architecture to facilitate the management of compute resources from different cloud providers in an homogenous manner. This preserves the flexibility and adaptability promised by the cloud computing paradigm, whilst enabling the benefits of cross-cloud resource utilization to be realized. The practical efficacy of the architecture is demonstrated through an implementation utilizing compute resources managed through different interfaces on the Amazon Elastic Compute Cloud (EC2) service. Additionally, we provide empirical results highlighting the performance differential of these different interfaces, and discuss the impact of this performance differential on efficiency and profitability.
Author(s): Dodda R, Smith C, van Moorsel A
Publication type: Report
Publication status: Published
Series Title: School of Computing Science Technical Report Series
Year: 2009
Pages: 15
Print publication date: 01/03/2009
Source Publication Date: March 2009
Report Number: 1144
Institution: School of Computing Science, University of Newcastle upon Tyne
Place Published: Newcastle upon Tyne
URL: http://www.cs.ncl.ac.uk/publications/trs/papers/1144.pdf
