This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| TORQUE Resource Manager | |
|---|---|
| Name | TORQUE Resource Manager |
| Developer | Adaptive Computing, US Department of Energy, Sandia National Laboratories |
| Released | 2002 |
| Latest release version | (varies) |
| Operating system | Unix-like |
| License | Proprietary and open-source variants |
| Website | (project pages) |
TORQUE Resource Manager TORQUE Resource Manager is a distributed job scheduler and resource control system used to manage batch workloads on clusters and supercomputers. It mediates between users, compute nodes, and scheduling policies to allocate CPUs, memory, and other resources for scientific and engineering tasks. TORQUE traces its lineage to early batch systems and is deployed across research laboratories, universities, and high-performance computing centers.
TORQUE provides batch queuing, job submission, job monitoring, and resource abstraction for cluster environments. It integrates with cluster middleware and orchestration stacks from organizations such as Open Science Grid, National Energy Research Scientific Computing Center, Lawrence Livermore National Laboratory, Oak Ridge National Laboratory, and Argonne National Laboratory. TORQUE exposes commands and APIs compatible with contemporaneous systems developed at Lawrence Berkeley National Laboratory and interfaces with workload managers from Hewlett Packard Enterprise, IBM, and Cray, Inc. installations. Administrators use TORQUE to present a uniform resource model to users who submit compute jobs using client tools and portals supported by institutions like University of California, Berkeley and Massachusetts Institute of Technology.
TORQUE originated as an evolution of batch systems created in academic and national-laboratory contexts. Early milestones involved collaborations among engineers at Sandia National Laboratories, Los Alamos National Laboratory, and contractor developers affiliated with U.S. Department of Energy programs. Over successive releases, TORQUE incorporated features to support emerging cluster architectures deployed at centers such as Argonne National Laboratory and Oak Ridge National Laboratory. The project ecosystem involved third-party contributors from companies including Adaptive Computing and open-source communities that maintained forks and patches used by European Organization for Nuclear Research-adjacent projects and university consortia.
TORQUE follows a client-server architecture with distinct daemons and utilities. The central server daemon maintains job queues and resource state, while node-level daemons manage execution and reporting on compute resources typical of systems at Sandia National Laboratories and Lawrence Livermore National Laboratory. Client utilities allow submission and control of batch jobs from workstations at institutions like Princeton University, Stanford University, and University of Illinois Urbana-Champaign. Key components include the server, scheduler integration points, resource manager plugins, and logging subsystems adapted to monitoring frameworks used by National Center for Supercomputing Applications and Pittsburgh Supercomputing Center. TORQUE’s on-node component interacts with job prolog/epilog scripts and environment modules common at facilities such as Brookhaven National Laboratory.
TORQUE delegates complex scheduling policies to integrated schedulers and policy engines used by high-performance centers. Typical scheduler integrations include systems adopted by Hewlett Packard Enterprise, Slurm Workload Manager-adjacent deployments, and site-specific schedulers at Oak Ridge National Laboratory. Resource control supports reservations, fair-share accounting, and node exclusion mechanisms employed in production clusters at Lawrence Berkeley National Laboratory and Argonne National Laboratory. Administrators leverage TORQUE hooks to enforce site policies, interface with accounting systems used by National Institute of Standards and Technology and to manage heterogeneous resources including GPU nodes similar to those at NVIDIA-equipped centers.
TORQUE is deployed in diverse environments ranging from departmental clusters at University of Cambridge and University of Oxford to national facilities operated by DOE. Integration patterns include coupling TORQUE with authentication and identity services provisioned by institutions like Internet2 partners, storage systems common to European Grid Infrastructure, and batch portals used by XSEDE. Site integrators configure TORQUE to work with configuration management and orchestration tools from vendors such as Red Hat and Canonical (company), and with monitoring stacks used at CERN and major supercomputing centers.
TORQUE’s design targets mid-size to large clusters and has been scaled in production at centers such as Oak Ridge National Laboratory and Argonne National Laboratory. Performance tuning typically involves adjusting server threading, database backends, and node heartbeat intervals in deployments similar to those at National Energy Research Scientific Computing Center. Scalability considerations include queue management, job-array handling, and reducing contention when thousands of concurrent submissions originate from research groups at University of Washington and Caltech. Comparative studies and operational experience at sites like Lawrence Livermore National Laboratory informed enhancements to logging and job lifecycle handling.
Operational security for TORQUE deployments follows best practices adopted by federal laboratories and research institutions including Sandia National Laboratories and Los Alamos National Laboratory. Administrators integrate TORQUE with site-wide authentication, authorization, and accounting infrastructures consistent with policies used by Department of Energy facilities. Licensing historically involved a mix of open-source distributions and proprietary offerings from vendors including Adaptive Computing; institutional deployments at National Renewable Energy Laboratory and university clusters reflect varying license choices and support contracts.
Category:Job scheduling systems