High-Energy Physics Applications on the Grid - Grid Computing: Infrastructure, Service, and Applications - page 441

Tier2s), while the ATLAS-specii c catalogs orchestrate the overall data dis-
tribution and bookkeeping. LHCb uses LFC as a global i le catalog. In this
case several Tier1s have a full read-only replica, synchronized with the
central one using Oracle data streaming functionality (Oracle Streams:
the replication is performed at the back-end level).
Globally (including non-HEP applications) over 100 LFC instances are in
use on the EGEE infrastructure. The largest installations have more than
10 million entries. The evolution of this successful product was always
driven by HEP use cases, although inputs from other user communities
have been taken into account. During this evolution the product included
more and more sophisticated features both to boost performance (like bulk
operations for inserting and deleting entries) and to cover security needs
(integration with the EGEE security infrastructure, data encryption, etc.).
Other catalogs exist developed by different experiments. One example
is the AliEn catalog, which is at the center of the AliEn system (the ALICE
distributed system) [24]. In this case the catalog not only keeps location
information for data i les (and corresponding metadata attributes) but is
also used by several components of the system. The catalog also contains
the information of software installations available at different sites and
the output of all the jobs.
As the i nal example of the fruitful collaboration between HEP and
other sciences on catalogs, I choose the AMGA metadata catalogue (AMGA
stands for ARDA Metadata Grid Access [25]). This system, originally
developed by ARDA as a tool to validate the metadata interface in the
EGEE middleware, was used as a laboratory to investigate efi cient tech-
niques to provide robust and efi cient access to databases in a grid context.
AMGA is the basis of a few systems in the HEP world (most notably the
LHCb bookkeeping catalogue).
The AMGA system has been adopted by several applications in com-
pletely different domains (see, e.g., the topic of Abstracts of the 2nd User
Forum organized by EGEE in 2007 [ 26 ] ). Applications range from
Climatology to Multimedia. The application we use here as an example is
High-Throughput Screening in Drug Discovery. The i rst application in
this i eld is WISDOM [27], active on the EGEE infrastructure since 2005. In
2006, a new phase was started with the arrival of new collaborators (most
notably by Academia Sinica Taipei [28]) and with the start of a set of
campaigns against the H5N1 virus (bird l u).
The basic idea is to use the grid to perform collaborative screening
of potentially active chemical compounds (called ligands). This activity,
called docking, can be executed on the grid by assigning single combi-
nations of proteins and ligands to independent execution units. In order
to scale up this activity, a central repository is needed (to assign the
protein-ligand pairs, to store and display the results, and to implement
more complex workl ows). The choice for this system has been AMGA
( Figure 17.10 ). The decisive arguments in the choice were the performance