KEDA CRON Scaling

Schedule Kubernetes Horizontal Pod Autoscaling (HPA) using KEDA’s Cron trigger to automatically adjust replica counts at specific times. This approach helps you:

• Handle predictable traffic spikes (e.g., morning rush).
• Save resources by scaling down during idle periods.
• Automate capacity management for promotional events.

How KEDA Cron Triggers Work

A Cron trigger in KEDA allows you to:

1. Specify an IANA timezone for scheduling.
2. Define start and end times with standard CRON expressions.
3. Set a desiredReplicas count for the active window.

When the clock matches your start expression, KEDA scales the target Deployment up to desiredReplicas. After the end expression, it scales back down (respecting your minReplicaCount).

Common Use Cases

When to Use Time-Based Scaling

• Predictable Load Windows: Morning routines, lunch peaks, end-of-business-day reporting.
• Short-Lived Events: Flash sales, marketing campaigns.
• Cost Management: Turn off extra capacity when it isn’t needed.

24-Hour Scaling Graph

This graph illustrates:

• Baseline: Default replica count until 06:00.
• Scale-Up: Jump to desiredReplicas during the 6:00–20:00 window.
• Scale-Down: Return to baseline after 20:00.

Defining the Cron Trigger in KEDA

Add a Cron trigger under spec.triggers in your ScaledObject manifest:

triggers:
  - type: cron
    metadata:
      timezone: Asia/Kolkata       # IANA Time Zone Database
      start: "0 6 * * *"           # 06:00 daily
      end: "0 20 * * *"            # 20:00 daily
      desiredReplicas: "10"        # Target replicas during active window

Complete ScaledObject Example

Below is a full example combining the Cron trigger with a Deployment target. It scales my-deployment up to 10 replicas between 06:00 and 20:00 Asia/Kolkata time, then scales down to zero after a 5 minute cooldown.

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: cron-scaledobject
  namespace: default
spec:
  scaleTargetRef:
    name: my-deployment
  minReplicaCount: 0
  cooldownPeriod: 300            # 5 minutes before scaling down
  triggers:
    - type: cron
      metadata:
        timezone: Asia/Kolkata
        start: "0 6 * * *"
        end: "0 20 * * *"
        desiredReplicas: "10"

With this configuration, KEDA automatically aligns your application’s capacity to match scheduled demand patterns, improving performance and cost efficiency.

Links and References

• KEDA v2 Cron Trigger Documentation
• IANA Time Zone Database
• Kubernetes HPA Guide