Cloud computing is developing in a rapid pace recently. Its easily provisioning property and the new pay-as-you-go cost model attract developers from all sectors to consider migrating their applications to the cloud platform. In fact, robust and reliable cloud platforms are available commercially.
At the same time, the research community has always been finding the most efficient way to use their research grant. While high performance computing helps to solve many complex algorithms, this has always meant heavy IT investment on hardware / software and ongoing maintenance cost.
In this project, we look into the problem on how an academic research project can make use of the cloud computing platform to gain scalability; and what is the cost effectiveness of doing so. In order to achieve these objectives, we focus on an application, namely Portimizer, developed by the Department of Statistics and Actuarial Science of the University of Hong Kong.
We have re-designed the architecture of Portimizer to tap into the vastly available cloud computing resources. The major goal is to reduce the total time needed to perform the complex calculations of Portimizer by scaling out the algorithms to the resources on the cloud computing platform.
By performing a set of tests using the modified version of Portimizer over a real life cloud computing platform (i.e. Microsoft Azure Platform), we have observed the scalability and cost-effectiveness of using cloud computing platform to perform research tasks. We have also learned some insights on the types of research tasks that are more suitable to be deployed to the cloud computing platform.
For details, please see our publication at here.