Student research opportunities
Energy Efficiency Issues in HPC Clouds
Project Code: CECS_713
This project is available at the following levels:
Honours, Masters, PhD
Keywords:
High performance computing, cloud computing, green computing, virtualization
Supervisor:
Dr Peter StrazdinsOutline:
Cloud computing has emerged as a popular paradigm for hosting business computing services. Interest has also grown for clouds as an alternative to supercomputers for high performance scientific applications. At the same time, energy efficiency in ICT is of a growing concern.
A background project has investigated dynamic re-scheduling techniques to improve the utilization of a
virtualized heterogeneous cluster (the hardware infrastructure for a cloud). The resulting framework, ARRIVE-F, is available as open source, and can dynamically migrate (the virtual machines hosting) parallel HPC jobs to the best-suited sub-cluster.
Goals of this project
This project will investigate energy considerations for clouds hosting services for parallel scientific applications. Particular challenges lies in the fact that these applications involve many communicating virtual machines which must be treated as a unit. Issues include:
- extending the ARRIVE-F framework to handle energy (as well as time) considerations.
- developing energy models for scientific clouds.
- incorporating energy considerations into Service Level Agreements
Requirements/Prerequisites
(PhD scholar) First-class or upper second class Honours degree in computer science, or equivalent. Experience in high performance computing or cloud computing would be an advantage.
Student Gain
This project is part of a collaboration with Cloud Computing and Green ICT Project at the Centre for Distributed and High Performance Computing at the Umiversity of Sydney, led by Prof Albert Zomaya. Collaboration and infrastructure opportunities also lie with the NCI National Facility.
Background Literature
Ashar Baig, A Cloud Guide for HPC (White Paper), Insight Consulting, May 2010
Young Choon Lee and Albert Y. Zomaya, Energy efficient utilization of resources in cloud computing systems, J Supercomputing, March 2010
Muhammad Atif and Peter Strazdins, Adaptive Resource Remapping In Virtualized Environments - Framework, TR-CS-11-01, ANU Computer Science Technical Reports, May 2011
