Knative, the Serverless That Isn't Serverless
summary
Knative is presented as a platform that reproduces many serverless behaviors but isn’t true pay-per-use because it still needs a running Kubernetes cluster and its costs; the talk references Prime Video’s ~90% cost claim to illustrate when keeping fewer always-on resources can be cheaper. It covers the Knative components Knative Serving and Knative Eventing, demonstrates creating a Python service with the kn CLI that pushes images to Docker Hub, mentions Kafka support and scale-to/from-zero capability, and shows a live demo (which failed to start pods) to illustrate event-driven patterns and the cost tradeoffs.
AI-written summary · about
chapters (10)
- 0:00Welcome
- 1:28Introduction
- 3:34Serverless vs stateless
- 8:57Stateless pros and cons
- 13:34Handling stateless flow
- 15:20Serverless events
- 19:41Knative’s uniqueness
- 21:30Knative use cases
- 23:53Knative demo
- 29:57Closing thoughts
related talks
35mStop Calling Knative Serverless!
21mServerless computing using Knative Open Source
45mKubernetes Serverless with Knative
20mKnative: A Kubernetes Framework to manage Serverless Workloads
23mServerless with Knative
52mServerless with Knative
31mKnative: Is it for me?
27mUnleashing the Power of Serverless Using Knative