Knative Tutorial: Revisions, Request Routing and Serving Scale
Knative Serving adds an application-serving model around revisions and request routing. This tutorial follows a declared service into an immutable revision and the path requests use to reach it. You will distinguish traffic assignment from replica count and examine the conditions around scaling down or activating work. The model is different from merely creating a Kubernetes Service, so its resources and controller behavior need to be read in their own context.
Here's a popular Knative video — watch it, then summarize it with one click. Or paste any other YouTube link below.
— or paste your own YouTube link —
Want to keep the key points from a YouTube video? Paste the link below to turn it into structured notes.
Why Use Knative Tutorial: Revisions, Request Routing and Serving Scale
- Connect configuration changes to revisions
- Understand traffic assignments between versions
- Separate request routing from workload replicas
- Inspect activation and scaling behavior
- Identify required platform components
Frequently Asked Questions
Is a Knative Service the same as a core Kubernetes Service?
They are different resource types with different responsibilities. A Knative Serving Service coordinates serving configuration and routing under its platform model. A core Service provides a Kubernetes network access abstraction and does not inherently supply revision management.
What does a revision identify?
A revision represents a particular immutable serving configuration in the Knative model. It provides version context for execution and routing. Changing current configuration creates another relevant version rather than silently making every existing revision identical.
Does shifting traffic rebuild an image?
Traffic assignment chooses which relevant revisions receive requests; image preparation is a separate process. A routing change does not inherently compile source. Verify that the chosen revision already refers to the artifact intended for use.
Why can scaling from zero affect a request?
A workload with no active instances needs activation before useful processing occurs under its configuration. That can introduce startup behavior. Evaluate response timing and application startup rather than assuming the absence of running replicas has no user-visible effect.
What should a Serving exercise inspect?
Inspect the service, relevant revision and routing or readiness conditions, then send the intended request. Those observations connect declaration to use. A created custom object alone does not establish that all required platform components and paths work.