By Bright Staff | September 01, 2020 |
Bucknell University, a private liberal arts college located in Lewisburg, Pennsylvania, has modernized its approach to managing their high-performance computing (HPC) cluster and is now using Bright Cluster Manager to provide consistency, ease-of-use, and the ability to easily extend its on-premise cluster to the cloud.
In 2018, the University was awarded a grant from the National Science Foundation to implement a Science DMZ that would house the next generation of its HPC infrastructure. After a careful evaluation of cluster management options, Bucknell made the decision to install Bright Cluster Manager to help deploy the new HPC environment, manage the HPC servers, and to provide a way to extend the HPC environment into the cloud. The HPC system at Bucknell is used by a broad range of disciplines, including Biology, Electrical & Computer Engineering, Mathematics, Physics, and Management, among others. End-users leverage the system to further the research efforts of faculty and students, as well as in the classroom for instruction.
Prior to installing Bright, Bucknell had managed their HPC servers with a home-grown system developed internally. In our experience at Bright working with thousands of organizations, there are many challenges with this approach including on-going development & maintenance of the internally developed solution, the lack of a modern user interface that provides overall control of the cluster, downtime resulting from errors made when making changes or upgrades, and image drift amongst servers to name a few. So, when Bucknell made the decision to implement Bright, they automated these challenges away and also reaped a significant return on investment in the process.
For example, Bright Cluster Manager saves Bucknell considerable time when rolling out new compute resources. "All it takes is to simply set the new compute servers to network (PXE) boot and point the network card (MAC) address to the appropriate image that we want installed. Bright does the rest - fully automating the installation,” says Jeremy Dreese, Systems Integrator, at Bucknell University. "Bright Cluster Manager has also simplified the administration of Bucknell’s Slurm workload manager, helping to seamlessly manage partitions and configuration options, and ensuring that the configurations are consistent across the environment. And, when it comes time to upgrade our systems, we simply update the server image once and Bright pushes the changes out to every server in the cluster".
Bright Cluster Manager operates at the heart of Bucknell’s HPC environment, facilitating new software deployment, provisioning hardware, and providing a complete management view of―and control over―the entire cluster. Bucknell now employs a cluster management solution that provides intelligent automation, which increases workload throughput and frees up administrators and researchers to focus on higher-value activities. With Bright Cluster Manager, everything is managed through a single web interface, and Bright's out-of-the-box monitoring and health checking ensures that the cluster will continue to operate at peak efficiency throughout its life cycle.
Additionally, Bright Cluster Manager was selected so Bucknell could diversify the services offered by their HPC environment, including a pathway to extending their resources to the cloud. Bright Cluster Manager automates the process of extending an on-premise cluster to the public cloud, creating a hybrid cloud solution that is easy to deploy, provision, and manage.
Bright Cluster Manager offers value to administrators with all levels of HPC know-how, enabling them to more quickly and easily service their end user’s needs, increasing productivity and accelerating time to results for the entire organization.
To learn more about Bright Computing and our installation at Bucknell University, please contact us at email@example.com.