- Speaker: Eric Wohlstadter (The University of British Columbia)
- Location: UBC Vancouver Campus, KAIS 2020, 2332 Main Mall, Vancouver, BC
- Date & Time: April 17, 2013, at 3:30 p.m.
Abstract
“Hybrid cloud” deployment can be an attractive option for companies wanting to deploy software services on scalable public cloud infrastructure, while still assuming local control over sensitive data resources. A hybrid deployment, despite providing better control, is difficult to design since code and data must be partitioned and distributed efficiently between a public cloud and private on-premise infrastructure.
In this talk, I will describe research into semi-automated partitioning of software services for hybrid clouds. The talk will address technical challenges relevant to: (i) modeling of software execution to abstract properties salient to the hybrid partitioning problem, (ii) combining code execution models and data storage models in order to support simultaneous optimization of execution data-flow and persistent data placement, (iii) formulation of the models to use integer programming for optimization. We implemented a new software profiling and partitioning framework (called Manticore) to evaluate our proposed solutions. Experiments on two open-source Web application show that Manticore can help developers to determine more optimal cost/performance tradeoffs for hybrid deployment than existing approaches.
This project is joint research with Nima Kaviani of the Computer Science Department and Rodger Lea of the Media and Graphics Interdisciplinary Centre (MAGIC).
Speaker’s Biography
Eric Wohlstadter is an Associate Professor of Computer Science at the University of British Columbia. His research interests are in the areas of software architecture, Web-based systems, cloud computing, and aspect-oriented programming.
