BNL RHIC Computing Facility

BNL RHIC Computing Facility
Name	RHIC Computing Facility
Location	Upton, Brookhaven National Laboratory
Established	2000s
Type	Research computing center
Affiliation	Brookhaven National Laboratory

BNL RHIC Computing Facility is the principal high-performance computing center supporting data processing for the Relativistic Heavy Ion Collider and related experiments at Brookhaven National Laboratory. It provides storage, networking, and compute resources that enable reconstruction, simulation, and analysis for collaborations including STAR, PHENIX, and other international projects. The facility acts as a regional hub linking to national and international grids and archives operated by institutions such as Fermilab, Lawrence Berkeley National Laboratory, and CERN.

Overview

The facility supports large-scale workflows produced by detector systems like STAR and PHENIX, coordinating with accelerator operations from RHIC and experimental schedules managed by Brookhaven National Laboratory. It interfaces with national programs including the Department of Energy-sponsored computing initiatives and collaborates with grid projects such as the Open Science Grid and WLCG partners. The RCF hosts compute clusters, disk arrays, tape libraries, and high-bandwidth network links to institutions including Stony Brook University, Yale University, Columbia University, and international centers like TRIUMF and INFN.

History and Development

The RCF originated to meet growing data volumes from RHIC commissioning in the late 1990s and early 2000s, aligning with milestones at Brookhaven National Laboratory and strategic plans from the U.S. Department of Energy Office of Science. Early partnerships involved computing groups from MIT, Princeton University, University of California, Berkeley, and University of Wisconsin–Madison, which contributed middleware and analysis frameworks. Over successive upgrade cycles, the RCF integrated technologies pioneered at Lawrence Livermore National Laboratory and Argonne National Laboratory, and adopted storage models influenced by Fermilab’s tape archive and CERN’s tape custodial strategies.

Architecture and Infrastructure

The RCF architecture combines commodity cluster nodes, high-memory servers, parallel file systems, and hierarchical storage managers. Compute nodes often use processors from manufacturers like Intel and AMD, and accelerators from NVIDIA for selected workflows. Parallel storage employs systems inspired by designs from Cray and NetApp, while tape backup uses libraries similar to those at Fermilab and CERN. Networking leverages connections to the ESnet backbone and regional optical links used by institutions such as Stony Brook University and New York University, with routing policies influenced by standards from Internet2.

Services and Operations

Operationally, the RCF provides batch scheduling, data cataloging, user support, and software distribution. Middleware stacks draw on projects like HTCondor, Globus Toolkit, and XRootD, with experiment-specific frameworks such as ROOT (software) and GEANT4 employed for analysis and simulation. User-facing services include interactive login nodes, web portals, and monitoring dashboards modeled after systems used at Oak Ridge National Laboratory and Lawrence Berkeley National Laboratory. Collaboration with experiment computing coordinators from STAR and PHENIX defines priority policies and allocation mechanisms.

Scientific Applications and Users

Primary scientific users are collaborations conducting heavy-ion physics, nuclear structure studies, and detector development—groups affiliated with institutions like Stony Brook University, Yale University, Columbia University, University of California, Berkeley, Massachusetts Institute of Technology, and Penn State University. Typical applications include event reconstruction, Monte Carlo simulation with PYTHIA and GEANT4, and data analysis in ROOT (software). The RCF also supports cross-disciplinary projects that interface with efforts at CERN, TRIUMF, and RIKEN.

Performance and Upgrades

Performance metrics have tracked network throughput, job throughput, and I/O rates, with periodic upgrades aligning with procurement cycles at Brookhaven National Laboratory and funding from the U.S. Department of Energy. Notable upgrade phases incorporated multi-core server fleets, SSD caching layers influenced by designs from Intel and Samsung, and expanded tape capacity modeled after Fermilab’s archival practices. The RCF participates in benchmarking against centers such as Oak Ridge Leadership Computing Facility and National Energy Research Scientific Computing Center.

Security and Data Management

Security posture follows federal standards and best practices used across national laboratories including Brookhaven National Laboratory, Argonne National Laboratory, and Lawrence Livermore National Laboratory, integrating identity management with directories and federations like InCommon. Data management combines metadata catalogs, checksum verification, and tape custodial policies similar to those at Fermilab and CERN, while compliance aligns with mandates from the U.S. Department of Energy. Incident response and vulnerability management collaborate with cybersecurity teams at Brookhaven National Laboratory and regional partners such as Stony Brook University.

Category:Brookhaven National Laboratory Category:Scientific computing centers Category:High-energy physics computing