PEGASUS: A Graceful Landing
Thu 15 Apr 2010
Back in 2006 the Engineering and Physical Science Research Council (EPSRC) funded a proposal to study how GridPP works. This project, PEGASUS, ran for almost four years and has just handed in its final reports to EPSRC. What has it been up to in that time? We caught up with Will Venters who is the lead researcher on the project at the London School of Economics to ask him a few questions about the project.
 Will Venters |
Whose idea was PEGASUS?
The idea of studying GridPP emerged out of pub-conversations with physicists and friends at UCL (in particular Peter Clarke and Mark Lancaster) and an MSc student I was supervising. She had intended to study Linux use by a commercial company, but at the last minute the company denied her access. She came to my office desperate for somebody to study and I thought “who do I know who use Linux and would be okay with her interviewing them”… A couple of weeks later she and I were interviewing Steve Lloyd and the rest is Pegasus!
Were physicists a group that always interested you from a sociological point of view or was it the particular system they were trying to implement?
It was both the systems and work practices which interested me. I had a good knowledge of how particle physicists approach problems (I am married to an ex-Particle Physicist) and think that the difference between their approaches and the commercial world (where most of my research has been) is fascinating. That said I have always researched distributed computing systems within distributed organizations so the Grid was a very good fit too. In essence it was a perfect match.
What was it like working with people in the grid world? Was anyone particularly forthright? Was it very different to the people you usually work with or study?
Yes it was different. The community is extremely open and happy to share its ideas. While it is taken for granted within GridPP this openness is something you seldom experience in the commercial world. This does mean forthright views, but these exist everywhere and it is better that they are said in public rather than hidden from view. One of the main challenges however was getting an understanding of the complexity of the work. We had to organize an afternoon class in particle physics from Prof Mark Lancaster to give the team a basic understanding of the field. We also have difficulties keeping up with the acronyms and terms used. Finally I genuinely found the people extremely nice and pleasant to be around. Liking beer helps!
Did any of your findings surprise or shock you?
Well firstly we are still making sense of the data – our findings will continue to come out in the next couple of years. That said we have produced a set of really interesting papers which outline our key findings. I guess the most shock has been the interest given by others to our work. Lots of people are interested in what we are doing and our papers are doing well.
Have you learned any lessons? Don't work with physicists?
I am keen to continue my work with GridPP even without the benefit of a project. In particular I think that the stress-testing of GridPP by LHC data will be a fascinating time. Pegasus had planned to study this originally but the LHC delays precluded this. Finally the LHC and GridPP are now an integral part of my teaching on Systems Development at the LSE – and the students comments suggest they find your work fascinating and valuable! As for lessons; the work challenges traditional assumptions about rationality and planning and gives a good example of how agile work practices can operate in a distributed settings. This concept of “Collaborative Agility” or “Scaled Agility” is new and valuable to others in other fields. It forms a core contribution from the project.
What do you think of the future of grids especially with the buzz around clouds and the formal formation of the European Grid Initiative earlier this year?
I think Cloud Computing is a very interesting area for Grid researchers. The level of interest within industry in Clouds is huge – mainly because it aligns with long-standing interests in outsourcing services. Further the increasing appreciation of the value of data-analytics within industries (even the Economist had a special issue on this – Feb 2010) will increasingly demand quicker and cheaper means to organize large-scale data analysis. I think Grid expertise could be useful in this. Certainly this is where my research on collaborative distributed work practices and computer systems is leading me.
Physics is renowned for acronyms even when they have to be forced. How long did it take you come up with PEGASUS? (Particle Physics Engagement with the Grid, A Socio-technical Usability Study)
Quite a while – probably a huge proportion of the time I spent on the grant application!
Now that PEGASUS is over what is your next project about?
I am currently doing some commercial work on Cloud Computing (writing a white-paper for a consultancy company) and am using GridPP as a case study within that. I am also involved in evaluating the Connecting For Health Electronic Prescribing Service with a particular focus on the distributed nature of that system. All this work follows from Pegasus and links to it. Finally I continue to write papers out of the Pegasus data – and will continue to hang around GridPP for a little while if you let me! I am also writing a Blog on Utility Computing drawing on my Grid lessons (www.utilitycomputing.wordpress.com) and have been including Pegasus outputs in my teaching!
More details on the PEGASUS project can be found at http://www.pegasus.lse.ac.uk
A full list of the papers generated by the project can be seen here: http://www.pegasus.lse.ac.uk/research.htm
© Copyright GridPP
If you wish to reproduce this piece please credit GridPP and contact Neasan O'Neill to say you are using it