Events — Colloquia & Seminars
Flexible Computing with Virtual Machines
Speaker: Andres Lagar-Cavilla, University of Toronto
Date: Tuesday, April 14, 2009
Talk: 1:30 PM, 366 WVH
Abstract
My research focuses on using virtualization to provide applications with new dimensions of flexibility in their execution. Specifically, I study flexibility along two axes: scale and location. Through scale flexibility, applications executing in the cloud can use a VM fork abstraction to instantaneously resize their computing footprint. VM fork statefully and swiftly replicates a VM to multiple hosts in parallel. VM fork provides well-known semantics and allow simple programming of parallel applications, flash crowd handling, load-balancing servers, etc. Through location flexibility, unmodified applications written in any language can be transparently and seamlessly migrated across the wide area to best match their resource needs. When reading data, applications are moved closer to the dataset; when going CPU-intensive, applications are moved to a remote cloud or supercomputer; and when the user is engaged with an interactive GUI, the application is moved to the user's site.
Brief Biography
none provided