Tuesday, November 13, 2018 – SC18, Dallas, Texas – Today, Bright Computing, a global leader in cluster and cloud infrastructure automation software, announced that Binghamton University in New York has upgraded to Bright Cluster Manager 8.1 to manage its clustered supercomputing environment. The project was managed by Bright Advanced Partner, Data In Science Technologies.
Binghamton University is one of America's premier public universities, attracting outstanding students and award-winning faculty who are partners in active discovery. Binghamton’s high performance computing cluster, aptly name “Spiedie”, is a 3464 compute core & 512 Knights Landing core cluster housed in the Thomas J. Watson School of Engineering and Applied Science's data center, located at the Innovative Technology Complex. This research facility offers compute capabilities for researchers across Binghamton University.
Since the deployment of the Spiedie cluster, it has gone through various expansions and deployments, growing from 32 compute nodes to its current 163 compute nodes.
Binghamton approached Data In Science Technologies (DST) in 2018, with the goal of upgrading its supercomputing environment, which included a Bright-managed cluster, RedHat/CentOS and Slurm, as seamlessly and quickly as possible. The DST team rose to the challenge and was able to stand up a new head node instance in parallel to minimize downtime and mitigate any impact on service.
Binghamton University chose to continue using Bright Cluster Manager as its clustered infrastructure management platform as a result of its ability to simplify the re-provisioning of compute nodes to new cluster images. Since deploying the latest version of Bright Cluster Manager, the IT team at Binghamton has found that Bright technology continues to be extremely easy to use, and day-to-day tasks remain simple to execute.
Phillip Valenta, System Administrator at Binghamton, Watson School of Engineering, commented; “Bright simplifies OS image management and deployment, so it’s easy to make changes of any size which are then pushed out to one node, or to all nodes. As our IT team is small, using Bright means we don’t require additional resources to support the backend of our HPC environment. In addition, some of our HPC applications can be installed and configured by Bright Cluster Manager repositories, which is a definite help.” He added; “I would recommend Bright Cluster Manager, especially for environments where personnel limitations exist. The streamlining of node image management and software implementation really reduces staff workload.”
Don Schrenk, Senior Partner at DST, commented; “Our strong relationship with Binghamton is based on us delivering a robust solution for their HPC environment, and we are delighted that the technology we have deployed is adding so much value to their productivity.”
The successful process of upgrading Bright in the Binghamton environment included:
- Inspecting system state, queues and health, backup OS, applications and settings
- Preparing environment for a parallel Bright environment on a virtual environment using Vsphere
- Installing new virtual image with Bright to v 8.x and deploy Slurm version 17.x
- Developing functional CentOS 7.x image and deploy to selected test nodes in Binghamton 163 compute nodes environment
- Recompilation, installation and test of applications in the new environment for CentOS 7, testing the Slurm queue using applications on a CentOS 7.x image
In addition, DST provided key ancillary services to further optimize the Binghamton HPC ecosystem.
- Integration and documentation of Bright Cluster Manager with Singularity
- Maintain existing SLURMDBD accounting database for past job data
- Maintain existing MYSQL databases for XDMod
- Compile latest OpenMPI against Intel Compiler
- Compile key applications for optimization against CUDA, KNL and CPU, Lammps, Gaussian, VASP
Clemens Engler, Director Alliances at Bright Computing, added, “It is often a challenge for University IT teams to deliver a quality HPC service while adhering to budgetary or staff limitations. Bright removes a lot of manual administration, freeing the Binghamton IT team to focus on providing an even better service to their researchers.”
For more information, please email firstname.lastname@example.org.