What is AWS Parallel Computing Service?
AWS Parallel Computing Service (AWS PCS) is a managed service that makes it easier to run and scale high performance computing (HPC) workloads, and build scientific and engineering models on AWS using Slurm. Use AWS PCS to build compute clusters that integrate best in class AWS compute, storage, networking, and visualization. Run simulations or build scientific and engineering models. Streamline and simplify your cluster operations using built-in management and observability capabilities. Empower your users to focus on research and innovation by enabling them to run their applications and jobs in a familiar environment.
Key concepts
A cluster in AWS PCS has 1 or more queues, associated with at least 1 compute node group. Jobs are submitted to queues and run on EC2 instances defined by compute node groups. You can use these foundations to implement sophisticated HPC architectures.
Cluster
A cluster is a resource for managing resources and running workloads. A cluster is an AWS PCS resource that defines an assembly of compute, networking, storage, identity, and job scheduler configuration. You create a cluster by specifying which job scheduler you want to use (Slurm currently), what scheduler configuration you want, what service controller you want to manage the cluster, and in which VPC you want the cluster resources to be launched. The scheduler accepts and schedules jobs, and also launches the compute nodes (EC2 instances) that process those jobs.
Compute node group
A compute node group is a collection of compute nodes that AWS PCS uses to run jobs or provide interactive access to a cluster. When you define a compute node group, you specify common traits such as Amazon EC2 instance types, minimum and maximum instance count, target VPC subnets, Amazon Machine Image (AMI), purchase option, and custom launch configuration. AWS PCS uses these settings to efficiently launch, manage, and terminate compute nodes in a compute node group.
Queue
When you want to run a job on a specific cluster, you submit it to a particular queue (also sometimes called a partition). The job remains in the queue until AWS PCS schedules it to run on a compute node group. You associate one or more compute node groups with each queue. A queue is required to schedule and execute jobs on the underlying compute node group resources using various scheduling policies offered by the job scheduler. Users don’t submit jobs directly to a compute node or compute node group.
System administrator
A system administrator deploys, maintains, and operates a cluster. They can access AWS PCS through the AWS Management Console, AWS PCS API, and AWS SDK. They have access to specific clusters through SSH or AWS Systems Manager, where they can run administrative tasks, run jobs, manage data, and perform other shell-based activities. For more information, see AWS Systems Manager Documentation.
End user
An end user doesn't have day-to-day responsibility to deploy or operate a cluster. They use a terminal interface (such as SSH) to access cluster resources, run jobs, manage data, and perform other shell-based activities.