Mastering Knative Sequence: Build Event‑Driven Pipelines with 4 Real‑World Scenarios

What is Knative Sequence?

When a piece of data must pass through multiple processing stages, developers call it a pipeline. Starting with Knative Eventing 0.7, the Sequence Custom Resource Definition (CRD) provides native support for such pipelines. This article introduces the Sequence definition and four common usage scenarios.

Sequence Specification

apiVersion: messaging.knative.dev/v1alpha1
kind: Sequence
metadata:
  name: test
spec:
  channelTemplate:
    apiVersion: messaging.knative.dev/v1alpha1
    kind: InMemoryChannel
  steps:
    - ref:
        apiVersion: serving.knative.dev/v1alpha1
        kind: Service
        name: test
  reply:
    kind: Broker
    apiVersion: eventing.knative.dev/v1alpha1
    name: test

The spec consists of three parts:

steps : defines an ordered list of services; each step creates a Subscription.

channelTemplate : selects the concrete Channel implementation.

reply (optional): specifies where the result of the last step is sent.

Four Practical Scenarios

Direct Service Access

Event‑Driven Processing

Cascading Sequences

Broker/Trigger Integration

1. Direct Service Access

An event source sends events straight to a Sequence, which invokes a chain of Knative Services (first → second → third). The services simply log the received event.

apiVersion: serving.knative.dev/v1alpha1
kind: Service
metadata:
  name: first
spec:
  template:
    spec:
      containers:
        - image: registry.cn-hangzhou.aliyuncs.com/knative-sample/probable-summer:latest
          env:
            - name: STEP
              value: "0"
---
apiVersion: serving.knative.dev/v1alpha1
kind: Service
metadata:
  name: second
spec:
  template:
    spec:
      containers:
        - image: registry.cn-hangzhou.aliyuncs.com/knative-sample/probable-summer:latest
          env:
            - name: STEP
              value: "1"
---
apiVersion: serving.knative.dev/v1alpha1
kind: Service
metadata:
  name: third
spec:
  template:
    spec:
      containers:
        - image: registry.cn-hangzhou.aliyuncs.com/knative-sample/probable-summer:latest
          env:
            - name: STEP
              value: "2"

apiVersion: messaging.knative.dev/v1alpha1
kind: Sequence
metadata:
  name: sequence
spec:
  channelTemplate:
    apiVersion: messaging.knative.dev/v1alpha1
    kind: InMemoryChannel
  steps:
    - ref:
        apiVersion: serving.knative.dev/v1alpha1
        kind: Service
        name: first
    - ref:
        apiVersion: serving.knative.dev/v1alpha1
        kind: Service
        name: second
    - ref:
        apiVersion: serving.knative.dev/v1alpha1
        kind: Service
        name: third

The resulting flow is illustrated below:

2. Event‑Driven Processing

A CronJobSource emits a JSON payload every minute, which is routed to the Sequence. After the three services process the event, the final result is sent to an event-display service for visualization.

apiVersion: sources.eventing.knative.dev/v1alpha1
kind: CronJobSource
metadata:
  name: cronjob-source
spec:
  schedule: "*/1 * * * *"
  data: '{"message": "Hello world!"}'
  sink:
    apiVersion: messaging.knative.dev/v1alpha1
    kind: Sequence
    name: sequence

apiVersion: serving.knative.dev/v1alpha1
kind: Service
metadata:
  name: event-display
spec:
  template:
    spec:
      containers:
        - image: registry.cn-hangzhou.aliyuncs.com/knative-release/event_display:latest

3. Cascading Sequences

Sequences can be chained, allowing one Sequence to feed its output into another. This enables complex multi‑stage pipelines.

apiVersion: messaging.knative.dev/v1alpha1
kind: Sequence
metadata:
  name: first-sequence
spec:
  channelTemplate:
    apiVersion: messaging.knative.dev/v1alpha1
    kind: InMemoryChannel
  steps:
    - ref:
        apiVersion: serving.knative.dev/v1alpha1
        kind: Service
        name: first
    - ref:
        apiVersion: serving.knative.dev/v1alpha1
        kind: Service
        name: second
    - ref:
        apiVersion: serving.knative.dev/v1alpha1
        kind: Service
        name: third
  reply:
    kind: Sequence
    apiVersion: messaging.knative.dev/v1alpha1
    name: second-sequence

apiVersion: messaging.knative.dev/v1alpha1
kind: Sequence
metadata:
  name: second-sequence
spec:
  channelTemplate:
    apiVersion: messaging.knative.dev/v1alpha1
    kind: InMemoryChannel
  steps:
    - ref:
        apiVersion: serving.knative.dev/v1alpha1
        kind: Service
        name: fourth
    - ref:
        apiVersion: serving.knative.dev/v1alpha1
        kind: Service
        name: fifth
    - ref:
        apiVersion: serving.knative.dev/v1alpha1
        kind: Service
        name: sixth
  reply:
    kind: Service
    apiVersion: serving.knative.dev/v1alpha1
    name: event-display

4. Broker/Trigger Integration

A CronJobSource sends events to a default Broker. A Trigger routes those events to a Sequence, and another Trigger forwards the Sequence’s output to the event-display service.

apiVersion: sources.eventing.knative.dev/v1alpha1
kind: CronJobSource
metadata:
  name: cronjob-source
spec:
  schedule: "*/1 * * * *"
  data: '{"message": "Hello world!"}'
  sink:
    apiVersion: eventing.knative.dev/v1alpha1
    kind: Broker
    name: default

apiVersion: eventing.knative.dev/v1alpha1
kind: Trigger
metadata:
  name: sequence-trigger
spec:
  filter:
    sourceAndType:
      type: dev.knative.cronjob.event
  subscriber:
    ref:
      apiVersion: messaging.knative.dev/v1alpha1
      kind: Sequence
      name: sequence

apiVersion: eventing.knative.dev/v1alpha1
kind: Trigger
metadata:
  name: display-trigger
spec:
  filter:
    sourceAndType:
      type: samples.http.mod3
  subscriber:
    ref:
      apiVersion: serving.knative.dev/v1alpha1
      kind: Service
      name: event-display

Conclusion

The article introduced the Sequence CRD, explained its spec, and demonstrated four real‑world patterns for building event‑driven pipelines in Knative Eventing. By selecting the appropriate scenario, developers can efficiently orchestrate multi‑step event processing on Kubernetes‑based serverless platforms.