Towards Serverless As Commodity: A Case of Knative
summary
Compares Knative, AWS Lambda, OpenWhisk, OpenFaaS and Kubeless across seven concrete API/runtime contracts and shows what must be standard for serverless to become commodity: packaging as OCI images, a pull-based runtime vs platform push, application contracts using CloudEvents envelopes or HTTP/STDIO, execution supporting sync+async, platform-provided retries (surprisingly only Lambda and OpenFaaS offer retries), and autoscaling with scale-to-zero (Kubeless lacks it). Analysis uses docs, hands-on Lambda deployments and Knative contributions and reports these practical interoperability gaps and tradeoffs.
AI-written summary · about
chapters (6)
- 0:00Start
- 2:28Kubernetes
- 10:56Definition: The API Boundary between the runtime & application in/out
- 18:50Packaging
- 22:14Invocation
- 28:23Execution
related talks
11mKeynote: Containerful Serverless
53mServerless is the Abstraction We Deserve
01h 00mServerless Madness on Kubernetes
21mServerless computing using Knative Open Source
40mWelcome to Serverless 2.0
31mKnative, the Serverless That Isn't Serverless
20mKnative: A Kubernetes Framework to manage Serverless Workloads
41mIs Serverless the End of Kubernetes?
