Taking decisions while on the move
Farid Ould-Saada, a professor of particle physics at the University of Oslo and head of the NorduGrid collaboration behind ARC, attributes his group’s achievements to being a tight-knit, efficient project group that is capable of taking decisions while moving ahead at full speed.
Solving CERN’s data challenge
ARC gives researchers the ability to work with Tier-1 CERN data centres. “ARC is a computer program, but a special one,” explains Professor Ould-Saada. “It means ‘Advanced Resource Connector’. Advanced, because we think it is advanced. And Resource Connector because you have distributed computing and storage resources you need to orchestrate. Around the year 2000 we faced what we call the data challenge. We really had too much data. Advanced instruments like the Large Hadron Collider (LHC) were producing much more data than we could even imagine. We ran into kind of a crisis, which at the same time was very interesting for computer science. We had to figure out how to take all this data, put it somewhere, access it, and be able handle and analyse it – which is the goal of experiments. ARC was developed to help solve this problem.”
Developed on working weekends
This is a solution that many scientists can use. “In the beginning, only the Nordic Data Grid Facility, now NeIC, was using ARC, but now several Tier-1 centres have adopted it because it is efficient. Switzerland, the UK and Germany rely on ARC now, and France is starting. ARC has become renowned because it is a way to access High Performance Computing centres. China is joining too,’ adds Professor Ould-Saada.
One special thing about ARC is that it was developed using little funding, which was possible thanks to the many who donated their spare time. Currently there are only a few full-time developers. As the professor explains it, “The idea has been developed in coffee breaks and in what Norwegians call a ‘dugnad’ (a group work effort). This is really teamwork. In six months we produced a grid that was demonstrated at CERN before anybody else, before several European projects could come up with a solution.”
A small, efficient team is the key to success
How were you able to do this?
Farid Ould-Saada: “I think it is due to our bottom up-approach. Ideas are good, but to deploy them in a realistic way is even better. We were probably the only ones working directly with system developers and administrators at the High Performance Computing (HPC) centres, which are closed environments for security reasons. So when we developed the code we took all the complexity into account. ARC works and it is much simpler than anything else. Moreover, we are a loose collaboration. We talk to each other and discuss which way to go every day. That’s what you can do when you have a small team. Very efficient. I think what you could call latency time is very important. When you have a research-important idea you sometimes have to ignore the stop sign, which is possible when you don’t have a big administrative framework. You have to be able to follow ideas and take decisions while moving ahead at full speed. To sustain the speed and quality of
Nordic, cost-effective and competitive initiatives, it is essential that our funding agencies follow the trend, and take decisions while on the move. ”
Why do you do it?
"I decided to be a physicist when I was something like 13 in some lost place in Kabylie, Algeria. I became a particle physicist, and then went into experimental particle physics, trying to answer some interesting, sometimes crazy questions. But then you need tools. You need an accelerator to collide particles, you need a detector to detect particles and suddenly we need to do a software and computing experiment to deal with a data challenge." Professor Ould-Saada continues:
"I think I became interested because it was a challenge. And then I was interested because I was on the right track, and afterwards I was still interested because if this is a good solution for us, it can be a good solution for more things even closer to daily life."
The interview is published in NordForsk Magazine
Text: Linn Hoff Jensen
Photo: Terje Heiestad/NordForsk