Presentation: Control Theory In Container Orchestration
What You’ll Learn
- Find out some of the basic principles of observing systems, controller design, and PID controllers.
- Learn how to apply engineering control theory to container management.
- Listen about scaling controllers, using both first principles and proven designs from Kubernetes and Mesos.
Abstract
Containers offer 2 central promises: software as encapsulated and reproducible units. A key corollary of those promises is the ability to schedule and manage containers automatically. This session aims to give attendees a solid understanding of leveraging or building container orchestration using feedback signals.
We’ll apply engineering control theory to key container management scenarios. This will cover basic principles of observing systems, controller design, and PID controllers. In particular, we’ll dive into container scaling controllers, using both first principles and proven designs from Kubernetes and Mesos.
Are you talking of automation systems control theory?
It's about applying control theory as an existing engineering rather than accidentally reinventing control theory when you're building automation. I'm going to introduce the fundamentals of control theory, the terminology, and I'll step through some increasingly complex cases in orchestration. Starting with something very simple, maybe health checking container, and going out to focus on auto-scaling. Incrementally bring in more terminology and bring in more concepts, such as cold PID controllers. Concept-scenario, concept-scenario.
Who's the audience you are addressing? Poope that are building orchestration tools or those using them?
I'm not sure if I have a focus between those two. A lot of it is to people who are building orchestrators or heavily customising, which can go from customizing code to just very heavily leveraging settings. Auto-scaling is something I want to focus on because that's something that many people heavily customize based on particular servers, particular apps, even if they're not writing platform.
When you say 'first principles', what do you mean in the context of orchestration?
I have this little container I want to do something with it without talking about abstract things already existing. In the case of health checks I would want to talk about that in terms of just hearers first off, the kind of state machine of what's going on with this computer when you're managing health check.
What do you want someone to leave your talk with?
I want to leave with a better understanding of how we can build automation controllers.
Similar Talks
Linux Foundation's Project EVE: A Cloud-Native Edge Computing Platform
Co-founder, VP Product and Strategy @ZededaEdge & Member Board Of Directors for LF Edge @linuxfoundation
Roman Shaposhnik
License Compliance for Your Container Supply Chain
Open Source Engineer @VMware
Nisha Kumar
Observability in the SSC: Seeing Into Your Build System
Engineer @honeycombio
Ben Hartshorne
Evolution of Edge @Netflix
Engineering Leader @Netflix
Vasily Vlasov
Mistakes and Discoveries While Cultivating Ownership
Engineering Manager @Netflix in Cloud Infrastructure
Aaron Blohowiak
Optimizing Yourself: Neurodiversity in Tech
Consultant @Microsoft
Elizabeth Schneider
Monitoring and Tracing @Netflix Streaming Data Infrastructure
Architect & Engineer in Real Time Data Infrastructure Team @Netflix
Allen Wang
Future of Data Engineering
Distinguished Engineer @WePay
Chris Riccomini
Coding without Complexity
CEO/Cofounder @darklang