Information processing on the LHC may point the way to the future of the internet
Recent information on the staggering amount of data that will be churned out by the Large Hadron Collider on a daily basis has led to a rethink on how one might access that volume of information remotely.
At its peak the LHC should output about 5 gigabytes of information for every five seconds it is running, translating into an annual output of 15 million gigabytes.
This volume of information cannot be easily handles by the net in its current state, bandwidth being a major obstacle. CERN has devoted some time to this problem and is currently handling the load in a series of tiers.
The LHC Computing Grid, as it is known, comprises Tier 0, located at CERN and which uses masses of CPU to process the raw data and ship packets over a dedicated 10-gigabit line to facilities worldwide.
These destinations are Tier 1 sites. From there the information is broken down to smaller packets and shipped to smaller global computer networks, classed as Tier 2 where they are analysed by the relevant personnel.
The key to co-ordinating this data, which may be spread over several networks is an open-source software called “middleware”. CERN is using a program called Globus at the moment to keep their data in line.
Brett Venter
