AWS has introduced the managed Parallel Computing Service (AWS PCS) service. This service enables high-performance computing (HPC) clusters to be set up on the AWS cloud infrastructure, especially for data and compute-intensive scientific workloads.
According to the cloud giant, the new AWS PCS managed service allows end users to easily run their simulations within HPC clusters on AWS and virtually scale these clusters. Users can deploy the Slurm scheduler for this purpose, work in a familiar HPC environment, and get faster results without worrying about the underlying infrastructure.
More specifically, AWS PCS simplifies management of HPC environments running on public cloud infrastructure using the AWS Management Console, AWS SDK and AWS Command-Line Interface (AWS CLI).
Admins can create managed Slurm clusters configured with their specific computing power, storage configurations, identity, and task allocation preferences. The managed service uses the Slurm scheduler, a highly scalable and fault-tolerant task scheduler several HPC customers use to set up and orchestrate simulations.
In addition, the service allows users to easily log into AWS HPC clusters, manage their HPC tasks, use interactive software on virtual desktops, and access data. AWS PCS also makes it easy and fast to add workloads without having to port code.
Building on AWS Parallel Cluster
AWS PCS builds on the previous tool for deploying and managing HPC clusters in the cloud: 2018’s AWS ParallelCluster. ParallelCluster allows users to deploy and manage HPC clusters in the AWS cloud environment and quickly set up proof-of-concept and production HPC environments.
However, users of this service wanted a fully managed AWS service for HPC clusters, primarily to eliminate the operational work around building and managing HPC clusters. AWS PCS is now the result of this.
Target groups and pathways
As mentioned earlier, AWS PCS is especially suited for existing and emerging compute—or data-intensive scientific workloads. These include applications such as computational fluid dynamics, weather modeling, detailed analysis of natural elements, automated electronics design, and so-called “reservoir” simulations.
The AWS PCS managed service is now available in several AWS cloud regions.
Also read: Can Ivanti still lose its image as a security vulnerability?