SDSC Leverages Bright Cluster Manager in new ‘Expanse’ Supercomputer

    

We recently announced that the San Diego Supercomputer Center (SDSC) will be using Bright Cluster Manager to manage the facility’s newest supercomputer, called ‘Expanse’. Bright Cluster Manager will enable Expanse to balance and manage resource diversity across virtually all domains of their science and engineering users, maximizing resource utilization, and increasing workload efficiency for research scientists across the country and beyond.

Bright Cluster Manager will provide comprehensive management of the Expanse system, enabling SDSC to administer its HPC resources as a single entity, provisioning the hardware, operating systems, and workload managers from a single unified web interface. Bright will also provide SDSC with access to:

  • Bright Autoscaler, which will enable Expanse to operate as both an HPC cluster and a Kubernetes cluster, supporting heterogeneous computing environments that combine Expanse with remote instruments, data sources, and edge devices.
  • Cloud Bursting, which will support SDSC’s expansion into the public cloud, enabling users to access the vast computational and data resources via direct scheduler integration with Expanse.

Expanse is a response to the National Science Foundation (NSF) call for a system that will provide computing capacity for research workloads serving a diverse set of end users. It will provide 5 petaflops of peak performance with Dell EMC™ PowerEdge™ servers with AMD EPYC™ processors and NVIDIA® V100 GPUs. The system has been designed to provide high performance and data movement across a diversity of applications. SDSC’s workload includes everything from modest scale batch jobs of tens of cores, to several thousand cores, to high-throughput computing jobs from the Open Science Grid, which are characterized by tens of thousands of single-core jobs.

SDSC operates at the highest levels of availability and utilization, and Bright’s monitoring and health checking will ensure that the entire cluster will continue to operate at peak efficiency throughout its life cycle. In choosing Bright, SDSC will be able to spend less time on systems administration and more time helping users be productive.

Click here to read the full press release. To learn more about Bright Computing and Bright Cluster Manager, please contact us at info@brightcomputing.com or visit our website at www.brightcomputing.com