High performance computing (HPC) system owners can spend weeks or months researching, procuring, and assembling components to build HPC clusters to run their workloads. Understanding and managing the complexities of compute, storage, networking, and software requirements can be confusing and time-consuming, slowing innovation and results. Once these elements are identified, it takes further guesswork to optimally size the cluster while balancing job queue times, cost, and agility for an evolving set workloads, each with unique demands favoring different combinations of solution components.
Understanding these challenges commonly faced by HPC customers, Amazon Web Services (AWS) and Intel have partnered together to introduce AWS ParallelCluster as an Intel Select Solution. AWS ParallelCluster is the first cloud based product to be verified as an Intel Select Solution. AWS ParallelCluster is an AWS supported open source cluster management tool that makes it easy for scientists, researchers, and IT administrators to deploy, manage, and automatically scale HPC clusters in the AWS cloud. This new AWS ParallelCluster allows customers, even those unfamiliar with AWS, to seamlessly orchestrate essential solution components to make the most of the AWS on-demand infrastructure. Intel Select Solutions are solutions available through Intel partners that have been certified to meet high performance and capability standards using Intel-based technologies.
AWS ParallelCluster is now available as an Intel Select Solution that takes advantage of certified cluster configurations that have met vigorous standards for HPC environments running simulation and modeling workloads. This means that customers who use AWS ParallelCluster can eliminate uncertainties of infrastructure setup and have the assurance their cluster meets a high standard of performance and suitability for their workload demands. Additionally, by using AWS ParallelCluster as an Intel Select Solution, customers can not only accelerate the process of choosing an appropriate cluster architecture, but even first-time AWS account users can stand up that infrastructure within minutes. As workload requirements evolve, the solution enables customers to automatically scale cluster capabilities up and down to match their real-time demands. Taken together, AWS ParallelCluster and Intel Select Solutions save users valuable time that would otherwise be spent building, procuring, or refreshing HPC systems.
A number of critical components underlying AWS ParallelCluster are essential in delivering the rigorous performance standards required to qualify as an Intel Select Solution. Customers can choose from three separate families of Amazon EC2 instances (C5n, M5n, and R5n) fueled by 2nd Gen Intel Xeon Scalable processors with built-in AI acceleration and 100 Gbps networking bandwidth. Customers can use the Elastic Fabric Adapter (EFA) network interface designed for scaling HPC workloads to thousands of cores in the cloud. Customers can use Amazon FSx for Lustre, a fully managed high performance filesystem based on the popular open source Lustre project, that offers sub-millisecond latencies and scales to hundreds of gigabytes of per-second throughput and millions of IOPS.
To learn more about AWS ParallelCluster and how to use it as an Intel Select Solution, visit this page.