You are here: Home About the RACF

A Guided Tour of the RACF

by John S. De Stefano Jr last modified Dec 21, 2015 08:55 AM
Contributors: Michael Ernst, Tony Chan, Shigeki Misawa
A history of the organization and facility management.

The organization that is now the RACF began when the RHIC Computing Facility (RCF) was established at Brookhaven National Laboratory in 1997 to support the computing needs of the experiments (BRAHMS, PHENIX, PHOBOS and STAR) at the Relativistic Heavy Ion Collider (RHIC). The RCF was a full-service scientific computing facility, which provided the bulk of dedicated computer processing, storage, and analysis resources for the RHIC experiments, along with general computing services for RHIC users, such as electronic mail, web services, file back-up services, and document processing.

In the late 1990s, Brookhaven was selected to serve as the U.S. Tier 1 computing facility for the ATLAS experiment at the LHC (Large Hadron Collider) at CERN. The ATLAS Computing Facility (ACF) was established in 2000 to support the computing needs of U.S. collaborators in the ATLAS experiment, leveraging the established infrastructure and capabilities of the RCF, and resulting in the formation of the RACF. In addition to utilizing the existing resources at the RCF, the ACF added support for newer computing services required to support the compute model planned by the ATLAS experiment. These new services were built upon the ideas of a global computing grid that grew out of the academic research in grid computing.

The major components of the RACF are the 450,000 HEPSpec 2006 (HS06) unit processing farm (currently with 50,000 processing cores), the distributed and centralized disk storage farm (over 40 PB of on-line disk storage), the robotic tape storage silos (200 PB of storage), a 3,000 terabit-per-second internal network up to 200 gigabits per second of Wide Area Network (WAN) connectivity, and data grid and cloud computing software infrastructure. The hardware is a combination of commodity-based processing servers, enterprise-class UNIX servers, and highly-specialized mass storage systems, all connected by a high-speed network infrastructure.

HPC Network Overview
The BNL HPC Network.

Since its establishment, the RACF has grown to its current level of about 30 staff members. The combined RACF staff operates and manages, year-round, a heterogeneous, large-scale, multi-purpose facility, serving a worldwide community of about 2,500 (and growing) users, while continuously innovating and addressing the ever-changing computing requirements of its user base.

For a detailed description of RACF facility resources and services, please see this HEPiX 2015 BNL Site Report presentation (warning: graphic-intensive document, about 60 MB in size).

Some pictures of the facility are available here.

For users or prospective users of the RACF facility, information on using the facility can be found on the Getting Started web page.

Filed under: ,