The face behind the Grid

Share this on social media:

Topic tags: 

John Murphy profiles the director of the Center for Grid Technologies, Information Sciences Institute

The entertainment industry is used to double acts; Sonny and Cher, Tom and Jerry, Batman and Robin, and so on. Carl Kesselman of the Information Sciences Institute of the University of Southern California has become famous as part of a double act with Ian Foster of Argonne National Laboratory.

They are the coordinators of The Grid - an ambitious project to give scientists around the world access to huge computing resources over high-speed networks, like Web services on steroids.

Apart from the technical achievement of designing software that allows big computational problems to be split up and run on servers located anywhere on a network, Kesselman and Foster have also achieved the monumental job of getting scientists and engineers around the world to work together on the problem rather than having several incompatible systems running. But persuasion is one of Kesselman's strong suits.

The Grid has attracted a lot of attention as scientific research is requiring access to more computing power than a single institution can provide. It has also struck a resonance with big experiments where research teams around the world need to share access to data.

Ron Ohlander, deputy director of the ISI, said: 'Carl has the ability to get people to co-operate. You could say he was a good salesman, but then the people he is selling his ideas to are the sort that are going to kick the tyres, so he also has to have something to back it up and he has that - he is very strong technically.

'As a person he is a human dynamo. He has an enormous capacity to work and can combine getting the work done with all his travelling and meeting with people to persuade them to join him.

'He is a good dinner companion, but he always looks like there are 15 other things on his mind while he is talking.'

Kesselman grew up in Rome in upstate New York. His father was a civilian engineer at a US Air Force Research base and as a child visiting his father's workplace and seeing the huge banks of computers stimulated his interest in technology. Although he didn't know it then, they were researching parallel computing. His father built his own home computer - pretty exotic at the time.

He was inspired to study electrical engineering although his great interest was music and he considered going to music school. His third major at college was music, but he dropped it in his final year to graduate in electrical engineering and computer science.

He managed to combine music into his post-graduate work and spent a year studying concert hall acoustics at the University of Delft in the Netherlands, which became the subject of his first published papers. The interest in acoustics led him into geophysics and then to signal processing, which he studied at the University of Southern California.

The move to Los Angeles was significant and he decided to stay. He lives in a bohemian quarter of Santa Monica near the beach and is heavily into the local lifestyle. His neighbours include some of the big names of Hollywood and he regularly used to pass film crews making episodes of Baywatch. He later 'hooked up' with Steve Crocker at UCLA, who was one of the original developers of the ARPANET and wrote RFC 1 and who had known his father while working for the Department of Defense. Crocker introduced him to the Aerospace Corporation, which allowed him to work part time while completing his postgraduate qualification.

Aerospace introduced him to the ideas of logic programming and artificial intelligence and he was so taken he decided to switch from signal processing to computer architecture as his main subject and he switched to UCLA. He had to redo a lot of his course work because he had changed direction, so his PhD took longer than planned. His thesis was on performance measurement of parallel programs.

During this time he met up with a young researcher from Imperial College, London, who shared his interest in parallel computing. Both were getting frustrated with logic programming because they couldn't see it applying to any real world problems. This was the start of his collaboration with Ian Foster.

The two of them set about looking at ways of using networking technology to build very large and powerful systems, getting disparate machines to work on parts of a problem and then combine for the result rather than trying to write software than could run on multiple processors in parallel. He and Foster put together a demonstration of their ideas called I-Way and started getting more people interested in working with them This was to result in a suite of software called Globus, which forms the foundation of the Grid.

His research was impressive enough for him to gain a postdoctoral position at the prestigious California Institute of Technology, working with Mani Chandy, which put him into an environment where he could mix with the top computer scientists in the world.

He said: 'At Caltech you had the thrill of switching on the radio in the morning to hear which of the guys you know has won this year's Nobel prize, and you just had lunch with them.'

After five years there he decided to move to the Information Sciences Institute at USC because while he loved CalTech he felt that he would be more at home at an institution associated with networking.

'I moved to ISI because it had a long and distinguished history of building important systems, one of which was the Internet. I had the opportunity to work with Jon Postel for a little bit and getting to know him a little bit was a good experience. He was an amazing guy. If you had an appointment with him in his of- fice you had his complete attention, he wouldn't answer the phone or anything. In a way it was like getting to know Alexander Graham Bell. Postel is one of the few people who is responsible for a real phase shift, and changed the world. He did it in an open way and never made any money from it.

'The other nice thing about being at ISI is that you are surrounded by 80 or 90 really challenging computer scientists. If you have a problem in computer security or networking or something else you just walk down the hall and speak to someone. If you want to know how FTP worked you can talk to the person who actually developed that technology.'

Kesselman has received a lot of support from ISI, which has invested in a Center for Grid Technologies that Kesselman now runs. But lately he has attracted support from IBM, which has put cash into the project and is planning to actually build a grid.

Kesselman sees this as an opportunity to run his ideas past people who are thinking in practical terms about everything, how the Grid would be built, supported and managed as much as the theory. But his most important collaborator is still Ian Foster, who Kesselman describes as his 'partner in crime'. Having worked together for 15 years it is clear that they get on well. Kesselman said they get a lot of work done in restaurants and bars as well as in their offices.

Kesselman believes the secret of their success has been that they have always centred their work on solving real problems rather than just developing abstract ideas. In the early days they thought about how their ideas might be used by meteorologists to run huge weather simulations and other computing problems.

A significant development is the realisation that sharing data is just as important as sharing computing resources. The Grid has shifted towards the idea of scientists around the world working in similar areas being able to get access to huge depositories of experimental data, which can processed into meaningful information across the same network.

They discovered that the Large Hadron Collider project at Cern in Switzerland was looking for similar solution and they are not the only ones. Had they not been active in promoting the one Grid idea it is likely that several disparate groups around the world would have wasted time duplicating each other's work.

The interest in his work means he is in such demand for meetings and conferences that he struggles to actually get the work done. There is certainly an issue there,' he said coyly.

Despite this he still has time for his music. He decided a few years ago to rediscover his talent for the clarinet and has started taking lessons. Maybe he wants to go down as the Pied Piper of computing?

Curriculum Vitae

1982
University of Buffalo, New York. BA Computer Science, BSc Electrical Engineering
1984
University of Southern California. Masters of Science, Electrical Engineering
1989
Visiting scientist, Swedish Institute of Computer Science
1990
Assistant scientist, Argonne National Laboratory
1991
University of California at Los Angeles. PhD, Computer Science
1982-1991
Member of the technical staff, Computer Science Laboratory, The Aerospace Corporation
1991-1993
Senior research fellow, Computer Science Department, California Institute of Technology
1993-1996
Member of the Beckman Institute, California Institute of Technology
Since 1996
Visiting associate, Computer Science Department, California Institute of Technology
Since 1997
Research Associate Professor, University of Southern California
Since 2001
Director of the Center for Grid Technologies, Information Sciences Institute, University of Southern California