High-Energy Physics Applications on the Grid - Grid Computing: Infrastructure, Service, and Applications - page 423

• Federate different grids, providing an effective interoperability

layer for data processing and data movement (e.g., DQ project
in ATLAS).

As a matter of fact all experiment-specii c layers contain all the three
patterns, with different level of emphasis depending on the needs of the
experiments and the different phases of their evolution. All these projects
were effectively a continuous stimulus to progress (in the HEP commu-
nity itself and in the grid communities at large). At the same time, they
allowed the maximum usage of the resources available from the different
infrastructures, overcoming interoperability and instability problems
observed in the early stages.
The early feedback from the user community was a decisive factor in
helping the evolution of these complex technologies. The HEP community
devoted signii cant resources (see e.g., the ARDA project described in this
paper [ 9 ] ) to work in close contact with the middleware communities.
The activities in (close connection with) the experiments eventually
matured in a coordinated process across all the active grid communities.
We are observing a sort of relay between the middleware community on
the one side, the infrastructure on the other, and the applications, in
particular HEP, on the third side. Over the years, three main phases have
been observed. The i rst one had the main focus on the development of the
middleware, especially prototyped in the pre-LCG phase. The second
phase corresponded to the i rst years of LCG (and EGEE): the goal was
essentially to demonstrate (by building it) a worldwide computing
infrastructure. Progressively the focus went to a third phase where more
and more feedback (and innovative ideas) were coming from the user
communities.
I believe that either the role of the applications (HEP and others) will con-
tinue to be strengthened (via close collaboration) or the existing momentum
will eventually be redistributed across national and application-specii c
solutions with possible loss of coherence. The HEP community, especially
for the sociological strengths and its power of innovation mentioned at the
beginning, is the best guarantee to keep the coherence achieved in the last
few years.
In recent years, very interesting patterns of collaborations have been
observed across different applications. In all major cases HEP played an
important role. Initially the idea of several projects (notably EGEE) was to
have the applications “validating” their services (the infrastructure, the
middleware) by injecting user requirements and in using prototypes.
In this perspective, a “generic” grid will be validated by exposing it to
several user communities (the more the better), effectively covering more
and more use cases. It is one of the main successes of these projects to
demonstrate grid usage from several applications (e.g., the spectacular
usage rise observed in EGEE-2). The key point is, however, different: an