"Evolution of the Internet" Powers Massive LHC Grid

jbrodkin brings us a story about the development of the computer network supporting CERN's Large Hadron Collider, which will begin smashing particles into one another later this year. We've discussed some of the impressive capabilities of this network in the past. "Data will be gathered from the European Organization for Nuclear Research (CERN), which hosts the collider in France and Switzerland, and distributed to thousands of scientists throughout the world. One writer described the grid as a 'parallel Internet.' Ruth Pordes, executive director of the Open Science Grid, which oversees the US infrastructure for the LHC network, describes it as an 'evolution of the Internet.' New fiber-optic cables with special protocols will be used to move data from CERN to 11 Tier-1 sites around the globe, which in turn use standard Internet technologies to transfer the data to more than 150 Tier-2 centers. Worldwide, the LHC computing grid will be comprised of about 20,000 servers, primarily running the Linux operating system. Scientists at Tier-2 sites can access these servers remotely when running complex experiments based on LHC data, Pordes says. If scientists need a million CPU hours to run an experiment overnight, the distributed nature of the grid allows them to access that computing power from any part of the worldwide network"

Comments

But does it run...

[RiotingPacifist]

Oh wait ofc it does, youve basically got science which is fundamentally open source.
Then youve got a bunch of scientists who are fundamentally geeks
And its all being setup in Europe, which isnt as under the grip of MS

As a bonus
They need to ability to look back and explain all their analysis which means they have to see the source
It costs a hell of a lot to get the data so they dont want to loose any data anywhere.
They have a lot of results to analyse so they dont want to be waiting for the server to come back on-line.
Could they of gone with BSD? probably, but most science tools are developed for linux.

Some Realtime

[Roger W Moore]

Actually not all of it is offline. One of the things I have a research grant for is to develop a realtime remote farm for monitoring the detector. This is to catch subtle detector problems quickly before we end up collecting 2 weeks of useless data.

For the Tier 1 a significant fraction of the data is raw 'sensor' (we call it detector) data. This allows reconstruction program converts the data into physics objects like electrons, muons, jets etc.) to be rerun on the data once bugs in the initial reconstruction program have been fixed.