Ramping up the Computing Services for LHC Data Analysis

Michael ERNST
DESY

The world's largest scientific instrument, the Large
Hadron Collider (LHC) is currently being assembled.
When operational, several
petabytes of data will be generated every year for a period
of at least ten years. These data will be acquired
at rates up to nearly 2GB/s and will be analysed by
thousands of physicists worldwide. In order to exploit the
full discovery potential of the LHC, a worldwide Grid is
currently being deployed. As part of the commissioning
of this Grid, a series of service challenges is currently
being conducted, ramping up the service progressively.
These challenges address not only the need to distribute
data reliably between many sites around the world - not
in burst mode but 24x7 for essentially all of the
production lifetime of the machine, but also and much
more importantly meet the needs of the experiments for
all of their offline data processing.
 
In this talk I will summarize the requirements of the
LHC collaborations as they are stated in their Technical
Design Reports, provide an overview of the Grid Service
Architecture, present some results from Service Challenge 3
and will close with an outlook concerning the ramp-up until
September 30th - when the environment devoted to worldwide
LHC computing is turned into production.