Biography
I am studying for a PhD in the Large Scale Distributed Systems group (Distributed Software Engineering) at Imperial College. I work on the Ukairo project, which aims to provide application-specific path performance improvements at Internet-scale.
My research interests mostly lie in computing at the extremes; either very large scale or very high performance systems. My main subject areas include: networking, operating systems, distributed systems, HPC, and security. I have a recreational UNIX abuse problem and a rational fear of C++.
Background
I am originally from Blairgowrie in north-east Scotland. I initially studied at the University of Edinburgh, from which I received a BSc in Computer Science. During this time I spent a year working in the Network Security Group at Sun Microsystems in Santa Clara, California. After graduating I worked in HP Labs in Bristol as a research assistant in the Enterprise Systems and Software Laboratory. Before beginning my PhD I completed an MSc in High Performance Computing at the Edinburgh Parallel Computing Center, also at the University of Edinburgh.
I have had a fairly varied career in computing over the last decade or so: I've developed software that has run on tens of thousands of systems; hacked on research languages; deployed code on two generations of national supercomputers; researched cloud infrastructure before it had such a catchy name; managed systems whilst wearing both boiler suits and lab coats; acted as door security to keep overly enthusiastic executives out of oversubscribed server launches; hung out of windows both deploying and policing rogue wifi access points; and on occasion crashed systems worth more than most houses.
Projects
My PhD work is on the Ukairo project which investigates how Internet routing can be improved by a detour routing overlay network, and how applications can use this enhanced routing to improve their performance. So far we have managed to develop a system that can double the median bandwidth to the Amazon S3 storage services from globally distributed clients.
I currently manage my research group's experimental networking cluster. At present it consists of 25 nodes with high speed Internet connectivity and is used for realistic evaluation of distributed systems as well as Internet measurement projects.
Publications
Refereed papers
- "Vsys: A Programmable sudo", Sapan Bhatia, Giovanni Di Stasi, Thom Haddow, Andy Bavier, Steve Muir, and Larry Peterson, USENIX Annual Technical Conference (USENIX ATC), Portland, OR, USA, USENIX, 06/2011 [PDF version]
- "On the Feasibility of Bandwidth Detouring on the Internet", Thom Haddow, Sing Wang Ho, Jonathan Ledlie, Cristian Lumezanu, Moez Draief, and Peter Pietzuch, 12th Passive and Active Measurement Conference (PAM), Atlanta, GA, USA, Springer, 03/2011. [PDF version]
- "Deconstructing Internet Paths: An Approach for AS-Level Detour Route Discovery", Sing Wang Ho, Thom Haddow, Jonathan Ledlie, Moez Draief, and Peter Pietzuch, Proceedings of the 8th International Workshop on Peer-to-Peer Systems (IPTPS'09), Boston, MA, USA, 04/2009. [PDF version]
Talks and posters
- "Ukairo: Internet-Scale Bandwidth Detouring" [Poster] Ukairo overview poster (February 2011)
- "Ukairo: Internet-Scale Bandwidth Detouring" [Slides] Ukairo overview talk for local seminar (February 2011)
- "Ukairo: Scalable Detour Routing for the Masses" [Slides] Talk given at MSN'10 (July 2010)
- "Deconstructing Internet Paths: An Approach for AS-Level Detour Route Disovery" [Slides] Talk given for paper at IPTPS'09 (April 2009)
Miscellanea
- PlanetLab Vsys tuntap - The original version of VSYS tuntap support for PlanetLab, allowing nodes to control their virtual interface devices, was developed as part of the Ukairo project.
- PlanetLab node mapper - Tool quickly hacked up to plot PlanetLab node locations.
- London pub list - London is full of terrible pubs. This is a handy list of some of the better ones from the perspective of someone who likes interesting beer.