Abstract: This talk will use the seminal twelve-factor app essay as a guide to discuss the do’s and dont’s of building and running containers. Each factor gives us an opportunity to consider avoidable anti-patterns if you’re using containers to deploy and manage repeatable, reliable, and portable services.
Containers rose in popularity on an oft used metaphor: lightweight virtual machines. We have a robust understanding of the benefits of virtualized hardware as a method of efficient resource utilization. The idea of even more efficient resource utilization makes sense. Unfortunately it’s a problematic metaphor.
Containers represent a constrained set of capabilities compared to virtual machines in order to make fine-grained guarantees about resource constraints and process isolation. This is a good thing. There is overlap in ideal capabilities between VMs and containers but it isn’t complete. Newcomers to the container ecosystem begin with a “lightweight VM” understanding and fall victim to specific anti-patterns.
After this talk you’ll understand common pitfalls in containerization and how you can avoid them. This discussion is useful for developers who wish to gain greater understanding of the environment their applications are deployed to, as well as operators interested in the benefits of containers for their architecture.
Learning Outcomes: - I’d like the audience to understand common mistakes made when considering containers as a “lightweight VM” strategy. Examples will include volume mounts for stateless workloads, multi-process containers (for example, ssh in your container), administering images without automation, building different containers for different environments, logging to files on disk, and many more.
- This talk will not only be anti-patterns. I will encourage best practices which align to each of the factors. Relative beginners in containerization will leave this class with an understanding of how to build and run containers using best practices.
