John Murphy profiles Paul Messina, former director of the Center for Advanced Computing Research at CalTech
Some people are hailed for their discoveries or inventions, but those who made them possible are often forgotten. Paul Messina has not been forgotten. A distinguished computer scientist in his own right - but where he really made his mark is in getting things done. Those who give out research grants trust Messina, because he delivers.
Although Messina has retired from his 'day job' as director of the Center for Advanced Computing Research at the California Institute of Technology (CalTech) he is in as much demand as he ever was. Having Messina's name on a research proposal goes a long way with the funding bodies, which is why not a day goes by without someone trying to inveigle him into their next project.
Messina is a true computing frontiersman. While others were talking and theorising about large-scale parallel architectures it was Messina who made them a reality and made them available to help tackle real scientific problems that were beyond vector machines.
Vince McKoy, Professor of Chemistry at CalTech, said: 'Paul really had the vision to work with the people who had the applications, to make the case to funding agencies in order to make world class machines available in such an amount that they would have significant impact.
'One doesn't often find a lot of enthusiasm on the computer science side of the fence for people who say: I like this machine and I really want to learn how to use it but I don't want to get involved in the nitty gritty, because if I can do that successfully then I can have a major impact on my field. That is where Paul comes through shining.
'Paul is someone you can trust. When you meet people like Paul you think 'what is his agenda?' Is he just asking me to come aboard and help to row the boat? Or is he really interested in the well being of everyone? Paul gets consensus. He works with people and he brings a sense of confidence to the table.'
Messina was born in the middle of an Allied air raid in southern Italy during the summer of 1943. His parents had been living in Salerno but they moved a few miles inland to escape the bombing. He lived in Italy until he was six when his family moved to Guatemala in Central America to escape the aftermath of the War. His father was a physician but it took several years before he was accepted for a medical job in neighbouring Panama. He spent his childhood in a small village on the Caribbean coast where his father worked in a banana plantation hospital. The local school only went to the eighth grade so his parents decided to send him to school in the US. At the age of 14, he was sent on his own to a prep school in St Louis, Missouri.
He said: 'I was concerned at first as to whether I would be able to keep up academically because it was quite a demanding school. But in the end that turned out not to be a problem.'
He started to develop a strong interest in science and particularly mathematics. He started at College of Wooster in Ohio majoring in chemistry and despite getting top marks in the class decided that he really didn't want to be a chemist. He admits the decision to go to Wooster was almost random because he knew nothing about US colleges and just went where his prep school adviser sent him.
He had a relative who was a science student in Chicago so he asked him for advice. He decided to change his major to mathematics and moved on to graduate school in Cincinnati to do a PhD in mathematics.
He said: 'I first became interested in computers when I went back to Panama one summer. I used to buy books before getting on the banana boat and read them on the voyage. In the summer of about 1963, I bought a book about computers and it piqued my interest. That fall when I went to university I found they had acquired an IBM 1620 mainframe. They had bought it primarily to do their accounting but they only needed it for a few hours and I got to sign up for time slots and use it. I taught myself to program and did a senior thesis in statistics principally because I wanted to do some computation.'
His computer experience proved useful in getting summer jobs with Procter and Gamble, which is based in Cincinnati, where he first started addressing scientific problems with computation. Messina got married while at graduate school and it became obvious that he needed a proper job if he wanted to start a family. He first worked in the university computing centre. He rapidly rose to manager of academic computer services while completing his dissertation, but once this was done, he decided it was time to spread his wings a little. He had become very interested in numerical techniques and applied for a job at Argonne National Laboratory, which had a group there doing this work. The legendary James Wilkinson, who had worked with Alan Turing on pioneering work at the UK's National Physical Laboratory in the 1940, dominated the group. Messina became a friend of Wilkinson and today cites this relationship as a big influence in his life.
Messina was initially hired by Argonne for his ability to manage a computer centre rather than his ability at mathematics but eventually he worked his way around to managing the sub-routine library as well as consulting in the computer centre. Promotion was rapid.
Messina said: 'After I had been there for just four years, the head of the Applied Mathematics and computing division left to work at the Department of Energy headquarters and I was offered his job. Suddenly I was the supervisor of all these people who I had admired and were the reason I had gone to Argonne in the first place.' Messina had decided by the early 1980s that parallel computing was going to be important in doing scientific calculations so he started steering Argonne towards making software that would still be valid in a parallel system. This set the groundwork for Argonne's present world leading position in parallel scientific computing.
Messina organised a workshop on parallel computing in 1986 where he met Geoffrey Fox from CalTech, who was also a pioneer in this field. Fox heard Messina speak and after the meeting approached him about moving to CalTech. By this time Messina had come close to getting funding for a large-scale parallel computing project but it had been cut by the Federal Government. Messina went for a visit and decided it was the right place to be. He was impressed by the company he would be keeping, with one of his technicians having previously worked for Richard Feynman.
Messina said: 'There are a number of universities that have no members of the National Academy of Science or Engineering, or maybe one or two. At CalTech, one third of the faculty is a member of a National Academy. Many of them are very approachable and it is a pleasure collaborating with them.
'Things can happen very quickly there because it is a small place. The president and provost let me do whatever I wanted. After my first visit, Geoffrey Fox asked me what I required. They didn't have an opening, as they typically don't. Within 10 days they had created an opening for me and offered me exactly what I wanted. 'An Italian friend of mine brought some students to visit CalTech and as they were entering the campus he told them "this is the cathedral of science".'
His work included running a NASA-funded high-performance computing project at the Jet Propulsion Laboratory but the main thrust of his work was trying to demonstrate that parallel computing was the way forward. Much of his work at this point laid the foundation for the Grid technology, which is one of the hottest topics in scientific computing today. The important thing for Messina was actually to build a large-scale machine so that people could run applications on it, rather than abstract computer science, because that way the scientists with the applications would come around and bring the computer scientists with them.
He said: 'Many people at CalTech were unhappy that people in the world at large did not think that parallel computing was useful, it was far too complicated and you couldn't get good use out of a big machine. But we felt that we knew exactly how to do that.'
Charles Seitz at CalTech had come up with the idea of the Touchstone machine that originally had 64 processors but no applications. Messina talked to Seitz about making a much bigger system and the idea was born for a system with 5-600 processors. DARPA and Intel both supported the project and it became known as the Intel Touchstone Delta. Messina headed off to Washington with some colleagues and started persuading agencies to fund part of the project each. Such collaboration was unheard of at the time. He got some money from the NSF, some more from the Department of Energy, eventually 13 institutions and four federal agencies funded the project. He allocated time on the machine according to what percentage each agency had paid towards its cost.
A contract was signed with Intel in 1990 and the machine was built. It was the fastest computer in the world for about 18 months. Although he is modest about this project, it cemented Messina's reputation with funding agencies because he had delivered exactly what he said he would do.
Messina's own work from then on concentrated on I/O because he realised that the problem was not computation but in moving the data in and out of such complex structures quickly enough to make them useable. His life also changed in that he was increasingly brought in to large national projects because of his reputation for being able to manage them. For a year, he was Director of the Office of Advanced Simulation and Computing, Defense Programs, National Nuclear Security Administration, in the Department of Energy.
All this took away from his 'day job' doing research and he admits now that he was becoming increasingly tired from so many years of working evenings and weekends. Just as one project was coming to an end he would be approached to join another, this time even more important and more demanding.
At the beginning of 2002, he announced he was retiring from CalTech. Some have said he might have stayed if some of his colleagues had appreciated him a little more and many distinguished friends at CalTech miss him. Since then he has returned to Argonne as an adviser and is working at CERN advising the director general on computing. He has also developed a relationship with the University of Lecce in his native Italy and he hopes his retirement will include visiting this university to give some lectures. Although his retirement is busier than most people's working lives, he hopes that he will be able to ease the pace a little.
He said: 'I took my two present jobs on the condition that there was no responsibility involved whatsoever. I am there to help and if people want my advice I am willing to give it.'
He is planning to spend more time sailing on his boat as well as travelling for holidays rather than work. As he watches the Grid take shape and parallel technology being at the forefront of scientific computing today he is laid back about the fact that he was doing all this 12 years ago. Those who are promoting the modern Grid all give deference to him as a major influence and even today in retirement there is a queue at his door of people who want the name Paul Messina on the title page of their budget submissions.
Manager, Academic Computer Services, and Adjunct Assistant Professor of Engineering Analysis, University of Cincinnati
Computer Scientist rising to Associate Director, Applied Mathematics Division and Head, Applied Mathematical Science Section, Argonne National Laboratory
Director, Mathematics and Computer Science Division, Argonne National Laboratory
Manager, High-Performance Computing, Jet Propulsion Laboratory
Director, Caltech Concurrent Supercomputing Facilities, California Institute of Technology
Director, Center for Advanced Computing Research, Assistant VP for Scientific Computing, Faculty Associate for Scientific Computing, California Institute of Technology
2002 to present
Adviser to Director General CERN, Distinguished Senior Computer Scientist at Argonne National Laboratory