The pyramid builder of our modern age
No job in physics at the moment can have a higher profile than leader of the IT division at Cern, the European Laboratory for Particle Physics, near Geneva, Switzerland. The large Hadron Collider (LHC) project, currently under construction there, is the biggest thing ever in Big Physics and is to scientific computing what building the pyramids was to playing sandcastles.
Its detectors will produce one petabyte of raw data per second, which must be filtered into one petabyte of useful data per year and distributed to experimental physicists throughout the world. The current design would require 200,000 of today's most powerful PCs. Its very feasibility depends upon the assumption that Moore's Law of computing power doubling every 18 months will continue to be valid until the start date in 2005.
Manuel Delfino has been in that leader's job since 1999. He has a rare combination of skills: he has been an experimental highenergy physicist; he has built detectors; and he has a long track record of applying computers to the output of those detectors.
In that time he has added to his skill set the ability to juggle the needs and interests of thousands of fellow physicists, together with the management of Cern and its participating member countries. This in itself is a Herculean task; after all, the pyramid builders had only one master. Delfino has proved up to the job, but he admits that even he cannot keep doing it forever, no matter how much prestige it carries. He wants to get back to doing science
David Stickland, a high-energy researcher from Princeton and computing project manager for the Compact Muon Spectrometer (CMS), said that Delfino is not a diplomat in the traditional sense but he is a good leader.
He said: 'The first thing about Manuel is that he is an extremely clever man who can think very quickly on his feet. He can point out exactly what is wrong with the argument you have just made even as you are finishing it. 'He is in a difficult position because he has many masters. He is a forthright person, which is not typical of a diplomat, but finds putting forward a well thought-out point of view as effective as beating people over the head.
'He looks just like a typical graduate student and carries his papers around in a rucksack rather than a briefcase. I have noticed that since he has been in his job he has started wearing a tie, which must help somehow. I reckon than when he returns to Barcelona the rucksack will stay but the tie will go.'
Delfino was born and brought up in Venezuela and he developed an early interest in the physical sciences from his father who was an engineer in the paper industry. His parents were comfortably off so they were able to send him to the US for his university education, hoping he would become an engineer.
He started at the University of Wisconsin, which allowed him to stay with relatives while he studied. He soon started taking more of an interest in physics than engineering and, while studying, got himself a part-time job helping with accelerator experiments.
He said: 'I used to work the night shift doing simple tasks. A lot of the operations were manual or with very simple automation, so I asked if I could have a computer account and used some of my time to create programmes that would automate some of my work. Eventually I worked my way out of the job.'
Fixing the computing for accelerator experiments was to become a theme of his life's work. His contacts allowed him to stay at Wisconsin as a graduate student and he eventually gained a PhD. in high-energy physics, which included some work at Fermilab and led to his first job as a post-doctoral research fellow. His work included developing a gas calorimeter for the MAC detector at the Stanford Linear Accelerator. Eventually while at Stanford, he met a professor from the Barcelona University called Enrique Fernandez. At the time the academic community in Spain was undergoing a huge renaissance following the restoration of democracy after the Franco era. There was already a high-energy research team in Madrid, but Fernandez persuaded Delfino to join him in setting up another group in Barcelona. In 1987 he moved with his wife, who is a social anthropologist, and now he loves the place.
He said: 'My wife and I both enjoy the urban lifestyle and going out to clubs. I am very grateful to Stelios from EasyJet who has brought down the price of travelling from Geneva to Barcelona, so I can get back there fairly regularly.'
His work at Barcelona was mainly concentrated on the Aleph detector at Cern and eventually he obtained a grant to work for a year at Cern on the computing needs of the detector, after which he returned to Barcelona as a research professor. He took a sabbatical from that job in 1994 to work at Florida State University, which was wrestling with the massive computing issues of handling the data generated by Cern detectors. Traditionally, these had been processed by mainframe computers but Delfino had earlier been working with emulator boards, which acted as sub-processors for the mainframe, allowing very high rates of data to be processed. In Florida he started working on the idea of using large clusters of cheap PCs to do a similar task and eventually was able to outperform Cern's mainframes. This was the start of the computing model he has proposed for the LHC.
After a year he returned to Barcelona as a full professor and started working on the Atlas detector for the LHC. The Atlas detector is the size of a six-storey building and Barcelona had won a contract to deliver a significant part of it. Meanwhile, his fame was spreading and he started being invited to speak at conferences on the issues of handling data from detectors. He also became chairman of the committee that was looking at the computing for the LHC.
He said: 'In the early days, I would write a paper and submit it to a conference and maybe they would allow me to present it. Things had changed and now I was being invited to speak at conferences before I had even started writing the paper I was going to present.'
He had always been active in Cern committees and in 1999 the new Cern director general, Luciano Maiani, offered him the job of leader of the IT division, putting him on both the main management and the main scientific committees.
He said: 'I am one of the people who walk the line between computing and physics. We may not know everything about each subject, but we can ask the right questions. However, if we knew what we were doing, we wouldn't be doing it. Not only is the task that we have to do impossible, we can prove that it is impossible. We have to get a little closer before we can work out how it can be done.'
The project is relying on the computer industry to supply the raw computing power and global research into Grid computing to supply the infrastructure for distributing the data. Delfino believes that the LHC computing project is a conceptual leap that will create the same kind of pressure for technological advance that programmes such as the Apollo space programme generated in the past. This has started to attract the interest of commercial sponsors who are keen to back the research because of the commercial spin-offs that will come.
In the meantime Delfino has to manage a tower of Babel. He said: 'If we were six months away from the start, we would be much more focused and there would be less politics. We would have the water up to our necks. What I try to do is to get people to focus on particular real problems; at the moment it is the LHC computing Grid. Then I make it so that the water is up to their necks.
'I don't believe this is a job that you should do forever, and in fact this is one of the longest jobs I have ever had. When I came to this job my technical batteries were fully charged and over the past three years they have gradually depleted. What I need to do is recharge those batteries. I will stay until the end of the R&D phase and then hand it over to someone else to do the implementation phase. All of us working on this phase will be building our personal feelings into it and so it is better for a new leader to come in with no pre-conceived ideas.'
BS Applied Mathematics, Engineering and Physics University of Wisconsin
UMS Physics University of Wisconsin
PhD Physics University of Wisconsin
Postdoctoral Fellow University of Wisconsin
Postdoctoral Fellow Universitat Autònoma de Barcelona
Senior Research Fellow, Fundación Banco Exterior (Cern resident)
Professor of Research, Universitat Autònoma de Barcelona
Visiting Scientist, SCRI, Florida State University
Professor of Physics, Universitat Autònoma de Barcelona
Leader, Cern IT Division, Switzerland