Deploying clusters as a service with Bright Cluster Manager on Dell EMC vSAN Ready Nodes enables organizations to spin up clusters on demand to increase access to HPC resources.
In recent years, high performance computing workloads have evolved from a consistent and well‑known set of applications for modeling and simulation into a wide range of applications for artificial intelligence, machine learning and data analytics. While these new data-intensive applications are driving the transformation of today’s enterprises, they bring their own set of challenges for IT organizations.
In particular, some applications increase the demand for HPC resources while requiring changes to the nature of the underlying infrastructure to support emerging workloads and new types of HPC system users. Today’s HPC clusters need to be more dynamic and more capable of leveraging new processors and accelerators. They also need to be able to incorporate heterogeneous infrastructure, integrate with hybrid multi-cloud environments, extend HPC to the edge and support self-service HPC for a wide range of users.
Needs like these build a case for a flexible approach to the delivery of HPC resources — and that’s the idea behind a new Clusters as a Service (CaaS) solution incorporating software from Bright Computing and VMware running on systems from Dell Technologies. This virtualized HPC solution brings together Bright Cluster Manager® on Dell EMC vSAN Ready Nodes to give data-driven organizations the ability to easily spin up clusters to increase access to HPC resources that drive innovation.
The solution gives organizations:
- The agility to respond immediately to user demands, automatically reassigning resources based on policies and enabling quick provisioning of self‑serve resources for authorized users
- The ability to automatically build a cluster from bare metal — setting up networking, user directories, security, DNS and more in a matter of minutes
- The flexibility to manage resources across on‑premises, cloud and edge environments
- Cost savings stemming from increased resource utilization via virtualization and reduced administration costs, hardware costs and public cloud usage
- The peace of mind that comes with a solution that is engineering‑validated, commercially supported and enterprise‑ready.
The big picture
Let’s take a step back and look at the big picture here. In recent years, advances in VMware® vSphere® software and hardware virtualization support have closed the performance gap between bare‑metal and virtualized deployments. These advances have made virtual clusters built from virtual machines and containers viable for a growing set of HPC applications.
So the promise is there, for sure, but the reality can be something else. Running VMs and containers alongside bare‑metal servers raises the bar for most HPC shops, and certainly for the growing ranks of commercial HPC environments. These factors drive the need for a professional, supported cluster management software platform that spans hardware, software and consumption models for the new era of HPC.
And that’s Bright Cluster Manager® running on an engineering-validated Dell EMC vSAN infrastructure that’s built to drive virtualized HPC clusters. Put it all together and you’ve got a powerful solution for capitalizing on clusters as a service — and spreading the goodness of HPC to more users.
For details on the technical specifications for the CaaS solution, see the Dell Technologies solution brief “Run VMs and containers alongside bare‑metal servers” and download the architecture.