Kubernetes Autoscaling

by Aaron Ackerman | at Minnebar spring 2020 (canceled)

Autoscaling! You want it, I got it. This is the session I wish I could have attended 9 months ago.

A look into how autoscaling in Kubernetes works using open source components. If it makes sense for you to be using Kubernetes it also makes sense for you to need automated scaling based metrics. I'll cover several different components that are important to autoscaling including: cluster instance scaling with cluster-autoscaler, metrics with prometheus, pod replica scaling with horizontal pod autoscalers using the prometheus metrics adapter, pod resource scaling with the vertical pod autoscaler, and automated instance draining for scale in using an external service (Lambda in this session).

Whatever cloud you choose don't be worried about getting something out of this. I guarantee 99% of the content will be cloud agnostic.

This session will start with an expectation that you have at least a passing understanding of Kubernetes and its core concepts like nodes/instances, pods, deployments. If you don't know about those concepts feel free to come anyways, you'll probably still learn something, but those concepts won't be explained in detail.


Aaron Ackerman

I am a senior site reliability engineer working on container orchestration. I tweet at @_aaronackerman_ on twitter and I write open source at @aackerman on github. Always striving to write high quality software.