What is KNative and how does it work?
Slide 2:
- Serverless does not mean without servers
- Server are completely abstracted
- Developers / Operators don’t have to think about resources (e.g. CPU and Memory)
- Scaled by Requests / Don’t pay for idling
- Your Code is the only artifact (Bring your own code)
Slide 3:
- Implementation to enable you to run serverless workloads on your cluster
- KNative is a framework that provides Custom Resource Definitions for Kubernetes
Slide 4:
- Serving
- CRDs for managing scale-to-zero deployments and routing to these application on top Istio
- These include a new Service definition that manages the deployment and the routing
- For every deployment a revision is created so you can jump back to an older state of your code in case you app is misbehaving
- Build
- Plugin-Framework to enable source-to-url workflows
- KNative Build is actually just able to run build in the cluster
- It’s up to you (or some 3rd Party) to provide mechanism for fetching code, creating and pushing images
- currently there are plugins to fetch from a git repository and to actually build using either buildpacks, jib or kaniko (build the Dockerfile in the repository root)
- Eventing
- plugin framework to react to specific events from different systems or to communicate to these systems
- Current implementations include a Kubernetes Source, a GithubSource and a GCP Pub/Sub Source