Containers Everywhere - HPC, Cloud and Edge
Friday, June 13, 2025 9:00 AM to 1:00 PM · 4 hr. (Europe/Berlin)
Hall Y2 - 2nd floor
Tutorial
HPC in the Cloud and HPC Containers
Information
Application software stacks are not only becoming more complex, but increasingly workflows are starting to extend single HPC systems. Containers have been identified as a key technology for addressing the challenge of deploying complex software stacks, whether the deployment happens on HPC, Cloud, or edge systems. With the emerging compute continuum, the ability to deploy containers in such a flexible manner has become critical. Examples of workflows extending HPC centres are weather forecast simulations, which run on HPC systems, that leverage input from observables collected from IoT devices. Another example is the efforts to make cities smart, where edge-controlled services are being coupled to compute-intensive machine-learning tasks.
Meanwhile, a good level of knowledge of using containers on HPC systems is available. However, deploying containers on different types of systems and designing containers with portability in mind remains a challenge for many. This tutorial addresses this challenge.
The tutorial focuses on key enabling technologies. Starting from different container technologies, we will introduce various technologies for launching containers on HPC, Cloud, or edge systems. More specifically, participants will learn about Docker and Singularity (Apptainer) containers and how to build these using a single source. Furthermore, Kubernetes, KubeEdge, and interLink will be introduced as technologies for steering the execution of containers on Cloud, edge, and HPC resources.
The goal of this tutorial is to create a basic understanding of key technologies and provide participants with opportunities to test these technologies based on relevant use cases. Another aspect that will be addressed is the challenge of designing containers in a portable manner, as the different types of compute resources are often based on different processing architectures.
To obtain hands-on experience, attendees of the tutorial will obtain access to a Kubernetes cluster, an HPC system, as well as edge devices.
Contributors:
Meanwhile, a good level of knowledge of using containers on HPC systems is available. However, deploying containers on different types of systems and designing containers with portability in mind remains a challenge for many. This tutorial addresses this challenge.
The tutorial focuses on key enabling technologies. Starting from different container technologies, we will introduce various technologies for launching containers on HPC, Cloud, or edge systems. More specifically, participants will learn about Docker and Singularity (Apptainer) containers and how to build these using a single source. Furthermore, Kubernetes, KubeEdge, and interLink will be introduced as technologies for steering the execution of containers on Cloud, edge, and HPC resources.
The goal of this tutorial is to create a basic understanding of key technologies and provide participants with opportunities to test these technologies based on relevant use cases. Another aspect that will be addressed is the challenge of designing containers in a portable manner, as the different types of compute resources are often based on different processing architectures.
To obtain hands-on experience, attendees of the tutorial will obtain access to a Kubernetes cluster, an HPC system, as well as edge devices.
Contributors:
Format
On Site
Targeted Audience
This tutorial targets developers, who are interested in the ability to use containers for designing and deploying workflows with components running on different types of computing devices. The tutorial is also of interest to administrators who want to improve their understanding of the presented key technologies and their practical use.
Beginner Level
50%
Intermediate Level
50%
