🎙️ Discussion

Containers Technologies for HPC Survey

Containers Technologies for HPC Survey

Welcome to the Container Technologies for High Performance Computing Survey!

Purpose

This survey is being conducted by the Containers Working Group with the goal of assessing the community’s opinion about container technologies for scientific computing. For more information, see the Survey Plan.

About Us

The Containers Working Group is hosted in the #containers-wg channel of the hpc-containers Slack workspace. We are a community of researchers interested in containers for High Performance Computing (HPC). Feel free to join the Slack and say hello!

This survey is completely optional, and you are free to stop at any time.

(This section needs more)

Questions

About Yourself (Demographics)

# Question Type Answers
A1 What is your primary environment? single National Lab, academia, commercial, private research institution, other
A2 What is your primary job? single Scientist, Linux Administrator, Research Software Engineer, Manager, other
A2.5 Which function is not listed in A2? text  
A3 How do you rate your experience with non-HPC containers? single no experience, beginner, intermediate, expert
A4 How do you rate your experience with HPC containers? single no experience, beginner, intermediate, expert

Container Runtimes

This section is focused to ask about HPC workloads running on HPC container runtimes. While some might use docker to run a single-node container on their workstation, we do not consider it in this set because it is not traditionally used in HPC environment.

# Question Type Answers
R1 Which HPC container runtimes are supported on the system(s) you are working on? multiple Charliecloud, Shifter, Singularity, Sarus, Podman, other
R1.5 Which HPC container runtimes are supported on the system(s) and not listed in R1? text  
R2 Which HPC container runtimes are you using for your daily workloads? multiple Charliecloud, Shifter, Singularity, Sarus, Podman, other
R2.5 Which HPC container runtimes are you using for your daily workloads which are not listed in R2? text  
R3 Which HPC container runtimes are on your radar but not yet used? multiple Charliecloud, Shifter, Singularity, Sarus, Podman, other
R3.5 Which HPC container runtimes are on your radar but not yet used and not listed in R3? text  
R4 Are you aware of the OCI runtime spec? single choice yes / no
R5 Do you care about the OCI runtime spec? single choice yes / no

Images

The image section aims to get an overview of how users are interacting with container images (building, sharing, Storing)

# Question Type Answers
I1 What is the method you are using to build a container image? multi Dockerfile, SIF recipe, From scratch, other
I1.5 What method did we miss to list in I1? text  
I2 Do you use higher-level abstractions on-top of the methods in I1? multi hpc-container-maker, repo2docker, spack, other
I2.5 What method did we miss to list in I2? text  
I3 Once build, do you push the image to a central registry? single yes / no
I4 What container registries are you pushing builds to? multi Docker Hub, Google Container Registry (gcr), Elastic Container Registry (ecr), Quay.io, GitHub Packages (ghcr,io), Azure

Potentials

Without making the survey too long, we could also ask the following questions.

  1. Do you have to build your container image for different CPU architectures?
  2. Do you care about available accelerators(GPU, FPGA, specialized network hardware) when building your container image?
  3. Do you use CI/CD for automated build and deploy?
  4. What are you biggest challenges or pain points when using containers, or reasons that you don’t use them (fill in the blank)
  5. What do containers not do that you wish they could? What would be an ideal future?