In the video Container Mythbusters, Michael Jennings presents some common myths about containers and explains the facts, including container technologies, the difference between containers and Virtual Machines (VM's), security vulnerabilities, and container runtimes for high-performance computing (HPC).
Containers may seem like a solution for portability when deploying applications on HPC platforms. However, deploying containers for scientific software on HPC leads to many questions, such as: Will building our application in a container enable it to run on particular HPC platforms? What are the vulnerability risks? Will this approach provide performance portability? Are containers a solution to reproducibility? Along with these questions and more, come discussions that can lead to assumptions or "myths" about containers. Michael Jennings explains the facts about container technologies for HPC in Container Mythbusters.
Some of the interesting topics in Container Mythbusters are:
- History from early attempts with chroot() to Docker and more
- What containers are (and are not)
- Privileged and unprivileged container runtimes
- Security vulnerabilities of privileged runtimes
- Docker is not the only container system
- Reproducibility challenges for computational and data science
- Explanation about user namespace for unprivileged container systems
This is a very informative talk about container technologies and uses for HPC. However, it does not provide any information on how to use any particular container solution.