The data grid, which will be one of the first places that IBM will test its so-called Storage Tank storage virtualization software, will house the information that CERN extracts from the Large Hadron Collider, a massive particle accelerator that can crash protons and electrons at sufficient energies to simulate the first moments after the Big Bang.
The Large Hadron Collider is being built in the Alps outside of Geneva, Switzerland on the border between France and Switzerland. It is an immense particle accelerator that, in theory, will be able to delve deeply into the nature of exotic (and so far only theoretical) particles, such as the Higgs boson, as well as the way that supersymmetry and antimatter contributed early in the life of the universe to the current configuration we live in. The LHC accelerator has a diameter of 27 kilometers and a large array of superconductors that need to be kept at a temperature of 300 degrees below zero. The LHC will use the same tunnel system created for an earlier accelerator, which was called the Large Electron Positron collide.
The data grid that IBM is building is only the beginning, with CERN expecting to generate several petabytes of data a year once LHC is running in 2007. Right now, CERN is starting out small with its data grid, having acquired six xSeries servers running Linux and 20TB of disk storage to test out the early versions of Storage Tank. IBM is contributing storage experts from it’s Almaden Research Center in San Jose, California, and Haifa Research Lab in Haifa, Israel, to extend Storage Tank. IBM Switzerland will support the whole shebang. As this equipment is ramped up and tested over the next few years, CERN will gradually increase the amount of storage in this data grid to 1 petabyte by 2005, well ahead of when the LHC itself is fired up.
The data grid project is being administered under the CERN OpenLab, a collaboration between CERN scientists and IT industry players like IBM, Hewlett Packard Co, Intel Corp, and Enterasys. HP and Intel are working with CERN to promote the use of Linux on Itanium servers. Last October, CERN joined the Gelato Federation, an organization established by HP and other technical computing centers in March to push Linux-on-Itanium. HP has bet its IT future on the Itanium processor, and it knows full well that it should use the momentum of Linux and other open source software to push Itanium platforms, which have significant benefits for high performance computing. HP and Enterasys, which supplies high speed networking equipment, are building a 32-node Linux cluster to initially support the LHC at CERN. Odds are that if CERN needs several petabytes of data storage, it will need hundreds of teraflops of processing capacity to chew through all that data and make sense of it. timpm@computerwire.com
Source: Computerwire
