SLAC at the LHC: Computing and Data Services
Not all of the Large Hadron Collider's hardware resides on the French/Swiss border; SLAC is playing a role in data analysis for the LHC's ATLAS detector right here on site. As a Tier 2 computing center, the lab has an integral part in the global effort to analyze the onslaught of data expected from ATLAS once the LHC comes online. SLAC is one of a few dozen Tier 2 institutions around the world, and one of only five in the U.S. At the moment, the main job of a Tier 2 institution is to run collision simulations. Once the LHC turns on, the lab's role will expand to include interpreting the data in search of a deeper understanding of physics at high energies.
"Intense data analysis is always a computing challenge," said Richard Mount, director of Scientific Computing Services and principal investigator for SLAC's Tier 2 project. "The unprecedented scale of BaBar physics analysis and our forward-looking work such as PetaCache, which involves massive random-access memory, have positioned us to create an agile and innovative environment for ATLAS analysis."
The tiered grid computing system enables the distribution and organization of the LHC's vast data and computing resources. CERN, the Tier 0 institution, will conduct the first round of data collection. The data will then find their way to one of the 10 Tier 1 institutions around the world, including Brookhaven National Laboratory—the only such site in the U.S.—for reconstruction of raw data into useful information such as particle energy and momentum. After multiple rounds of further Tier 1 refinement, the data are ready for Tier 2 computing centers like SLAC, which will provide researchers with the computing power they need to begin analysis for their particular projects.
"One of our biggest challenges is finding the best way to transfer this huge volume of data to and from the lab," said Wei Yang, who serves as the technical lead on the SLAC Tier 2 project. "The network pipe has to be fat."
To handle the huge data flow, SLAC network experts and the SLAC Tier 2 team—which consists of Booker Bense, Chuck Boeheim, Andy Hanushevsky, Randy Melen, Lance Nakata, Mount and Yang—upgraded the network infrastructure from one gigabit to 10. The team is also preparing high-performance disk space to hold the data, and a random-access system through which to access it.
"Right now, we constantly have jobs coming in, but those are mostly simulation jobs. When the real data start coming in, that's going to be the challenge," Yang said.
—Kelen Tuttle and Marcus Woo