High-Energy Physics Applications on the Grid - Grid Computing: Infrastructure, Service, and Applications - page 422

had to be planned well in advance both for reasons of cost and to
master the increasing complexity of the scientii c data. Along
these lines, CERN set up a review of the LHC computing (the so-
called Hoffmann review [6]) in 1999 to prepare and formally
secure the mechanisms to build and maintain the necessary com-
puting infrastructure. Eventually, the LHC Computing Grid
(LCG) project, led by Les Robertson (CERN), was started in 2001
[7]. Notably, the LCG project was designed with the necessity to
“cross” the experiment boundaries, fostering cross-experiment
collaborations at the level of base tools (both in the application
sector and in the infrastructure).

3. There were many examples of HEP experiments using distrib-
uted computing infrastructure well before LHC, notably on
national centers like the IN2P3 Computing Centre in Lyon (France)
or CINECA and CNAF in Bologna (Italy). The important point is
that the grid concept suggested a complete solution to concrete
issues being experienced in the HEP domain (single sign-on, role-
based access and global sharing of resources). When I. Foster
delivered a very inspiring talk at the CHEP (Computing in High-
Energy Physics conference, the lead event for computing in the
HEP community) in March 2000, the HEP community was already
designing (and validating with simulation studies and prototyp-
ing work) a hierarchical model, which is still the foundation of the
LCG infrastructure (MONARC project [8]).

4. The HEP community was at the heart of the European Data Grid
project (EDG), led by F. Gagliardi (CERN) who then initiated the
EGEE program. The HEP experience, together with innovative ideas
and tools from the grid community (most notably the Globus proj-
ect led by I. Foster and the Condor project led by M. Livny), initiated
a number of research and development studies on the middleware
necessary to provide dependable services for user communities
(HEP plus Biomedical and Earth Observation applications). The
software stack adopted and evolved in EDG and then in use by
EGEE is the underlying base for operating the grid. In parallel, sev-
eral initiatives have been undertaken by the experiments to provide
high-level services to serve specii c needs. All the LHC experiments
developed layers on top of the base grid services. The reasons were
multiple, but in general we recognize the following patterns:
Insulate the physicists community from an infrastructure that •

is in fast evolution (e.g., AliEN project developed in ALICE).
Provide a layer to optimize performances, in particular to •

increase efi ciency, stability and minimize latency (e.g., DIRAC
project developed in LHCb).