PBS Professional

PBS Professional
Name	PBS Professional
Developer	Altair Engineering
Released	1995
Programming language	C, C++, Python
Operating system	Linux, Unix
Genre	Job scheduler, Workload manager
License	Open source / Proprietary

PBS Professional

PBS Professional is a high-performance workload management and job scheduling system used to orchestrate compute jobs on clusters, supercomputers, and cloud environments. It coordinates resource allocation, job queuing, and execution across heterogeneous nodes, integrating with batch systems, parallel libraries, and resource managers. PBS Professional is employed in scientific research, engineering, and enterprise computing centers that require sophisticated scheduling and policy enforcement.

Overview

PBS Professional provides centralized control for batch job submission, priority policies, and resource reservations on compute infrastructures such as Oak Ridge National Laboratory, Lawrence Livermore National Laboratory, Argonne National Laboratory, National Energy Research Scientific Computing Center, and commercial data centers run by Amazon Web Services, Microsoft Azure, Google Cloud Platform. It interoperates with MPI implementations like OpenMPI and MPICH, storage systems such as Lustre and GPFS, and identity systems like LDAP and Kerberos. Administrators use PBS Professional alongside monitoring tools including Prometheus, Grafana, and Nagios to observe cluster health and job metrics.

History and Development

PBS Professional originated from research at the NASA Ames Research Center and early parallel computing projects supported by the United States Department of Energy. The software lineage ties to academic and national laboratory work that overlapped with projects at Sandia National Laboratories and Los Alamos National Laboratory. Commercial stewardship transitioned through companies that included Veridian Corporation and later Altair Engineering, which guided development to support exascale-class systems and cloud orchestration. Over time, PBS Professional integrated features inspired by schedulers like HTCondor and SLURM, while maintaining compatibility with batch workflows from projects associated with XSEDE and the Open Science Grid.

Architecture and Components

The architecture of PBS Professional centers on a server-agent model with components that include a central scheduler, a server daemon, and node-level execution daemons. The scheduler enforces policies and priorities analogous to mechanisms used in Cray system resource managers and coordinates with queueing semantics seen in SGE-derived systems. Components include job submission clients compatible with command-line tools used in GitLab CI pipelines and RESTful APIs patterned after services such as Kubernetes for cloud-native integration. Storage and network interplay mirror deployments integrating with high-performance fabrics from Mellanox Technologies and interconnects like InfiniBand.

Features and Functionality

PBS Professional implements advanced scheduling policies, fair-share accounting, preemption, backfilling, and reservation capabilities used by centers comparable to NERSC and Fermilab. It supports array jobs, job dependencies, and heterogeneous resource descriptions similar to features found in Torque (software) and OpenPBS. Security and authentication mechanisms integrate with standards exemplified by X.509 certificates and federated identity projects such as InCommon. Accounting and reporting tie into billing systems used by facilities like Argonne Leadership Computing Facility and job profiling tools like TAU (Tuning and Analysis Utilities) and Vampir.

Deployment and Use Cases

PBS Professional is deployed across academic institutions including University of California, Berkeley, Massachusetts Institute of Technology, and Stanford University as well as industry users in aerospace firms such as Boeing and automotive companies like Ford Motor Company. Scientific domains include computational chemistry projects associated with Gaussian (software), climate modeling linked to NCAR workflows, and genomics pipelines comparable to deployments at Broad Institute. Cloud adoption patterns parallel orchestration strategies used by OpenStack and container workflows involving Docker and Singularity.

Licensing and Governance

The product is available under mixed licensing models that have evolved alongside contributions from open-source communities and corporate stewards such as Altair Engineering. Governance reflects collaboration models seen in projects stewarded by organizations like the Apache Software Foundation and consortiums such as the Scientific Linux user groups, balancing community contributions with commercially supported offerings. Compliance, export controls, and procurement practices often mirror policies from agencies like the U.S. Department of Commerce and funding agencies such as the National Science Foundation.

Category:Job scheduling software Category:High-performance computing software