Kubernetes Requests, Limits, and Autoscalers: How They (Sometimes Don't) Work Together

Video

About the Talk

Did you know that setting the wrong requests and limits for pods can crash your entire cluster?

See a breakdown of the interconnected parts involved with automating a Kubernetes configuration, from defining requests and limits, to understanding the complex nature of autoscalers and schedulers.

Learn best practices on balancing these automated tools to assist rather than hinder your application infrastructure.

What You’ll Learn

• What are requests and limits? Why do we use them?
• How does the Cluster Autoscaler work?
• How does the Horizontal Pod Autoscaler work?
• How can you set intelligent requests and limits in YOUR cluster?

This Talk Is Designed For

New and experienced Kubernetes developers that want to learn more about how these complex Kubernetes services interact.

Resources

Slides

Kubernetes on DigitalOcean: Docs

Requests & Limits Crash Course by Henning Jacobs [Video]

CPU limits and aggressive throttling (The CFS bug he mentions has been patched.)

HPA Algorithm details

Using the Vertical Pod Autoscaler

Kubernetes best practices: resource requests and limits

Kubernetes best practices: configuring liveness probes

CPU Management [Kubernetes Documentation]