Distributed data analysis – grid computing

The experiments at the Large Hadron Collider at CERN collect data from particle collisions in order to study the fundamental particles and their interactions. The dataset collected amounts to several petabytes of information that must be stored and analysed in order to extract scientific results. When performing these calculations, computing resources from many different parts of the world are used, connected together to form one computing grid, the Worldwide LHC Computing Grid (WLCG).

Aims
In order to build a computing grid from geographically distributed resources a grid middleware is needed. The middleware provides a common interface for submitting computing tasks to different computing resources. It alse provides information about the different resources so that the tasks can be matched to a suitable resource. Our project is developing the grid middleware called the Advanced Resource Connector (ARC).

Methods
Compared to its competitors the ARC middleware is particularly useful for connecting non-dedicated and heterogenous computing resources, which has lead to that it is being used in an increasing number of sites, also in new locations such as China. The ARC middleware originated as a Nordic project, but the collaboration developing and maintaing it has grown to include members from many countries in Europe. The Uppsala team has contirbuted to various parts of the ARC development, especially the client library and the information system. We also maintain the installation packages of the ARC middleware in various Linux distributions such as Fedora, EPEL and Debian.

Research group

PI:Prof. Richard Brenner
Dept of Physics and Astronomy, Uppsala University
Dr. Mattias Ellert, Henrik Öhman
Dept of Physics and Astronomy, Uppsala University
Dr. Balázs Kónya, Dr. Oxana Smirnova
Dept of Physics, Lund University

Links and references