News

Endless Cloud double edged sword


29 August 2013

It is critical that we develop models for the efficient use of cloud resources and traditional data centres, says Dr Young Choon Lee
It is critical that we develop models for the efficient use of cloud resources and traditional data centres, says Dr Young Choon Lee

Hybrid or multi-cloud deployment is the most cost effective and efficient method for a business to manage abundant cloud computing resources, cloud computing experts at the University of Sydney say.

The School of Information Technologies' Centre for Distributed and High Performance Computing has endorsed the hybrid cloud model because it is a scalable, quantifiable solution that can combine the best of both private and public clouds. Their endorsement comes on the heels of the group's Cloud Computing Systems and Applications forum.

"A hybrid cloud is a cloud computing environment where a business can manage some resources in-house while also utilising external providers," explains Dr Young Choon Lee, a research fellow at the centre.

"A business or organisation might use a public cloud service such as Amazon EC2, MS Azure or Google Compute Engine to process and archive burly data, but continue to utilise its in-house storage capacity for operational or day to day data.

"The hybrid approach allows a business to take advantage of the scalability and cost-effectiveness that a public cloud computing environment offers without exposing the organisation's applications and data to third-party vulnerabilities such as hacking."

"The availability of virtually unlimited cloud resource with the elasticity of public clouds in particular is a great opportunity for a business yet determining the best resource provisioning level is an open question."

The trade-off between cost and performance that are two main and conflicting objectives in hybrid cloud can be effectively captured using pareto-optimality, states Dr Lee.

Corporations can choose multiple trade-off points best suited to their business needs

"For example, pharmaceutical companies could run their drug design software - software that typically requires large-scale distributed computing capacity - across private and public clouds."

Professor Albert Zomaya, Chair of High Performance Computing and Networking at the University who also participated in the think-tank says it is critical that we develop models for the efficient use of cloud resources and traditional data centres.

"Data centre usage can be below 20 percent. Under usage causes a number of issues including the ineffective use of energy and excessive costs."

Many organisations already combine their own private clouds with data centres. Managing their cloud storage addresses security concerns but not consumption or usage levels.

"Private cloud capacity can be much reduced by dynamically provisioning public cloud capacity. While sensitive data and time critical workloads are processed in their private cloud, the rest may be offloaded to public clouds."

While a seemingly cost effective option for excessive data, businesses must consider security and regulatory compliance requirements such as privacy and confidential of client information, states Professor Zomaya.


Follow University of Sydney Media on Twitter


Media enquiries: Victoria Hollick, 02 9351 2579, 0401 711 361
victoria.hollick@sydney.edu.au