Domain 1 of 4 · Chapter 5 of 7

Event-driven scaling with KEDA

Unlock the complete study guide + 1,040 practice questions across 16 full exams.

Bundled into the existing Developing AI Cloud Solutions on Azure premium course — no separate purchase.

14-day money-back guarantee — no questions asked.

Included in this chapter:

  • Limits, rules, and behavior: one scale definition
  • From zero to one, then from one to many
  • Three trigger categories, and what each measures
  • Translating a KEDA scaler into a scale rule
  • More than one rule on one app
  • Scale to zero, and the rules that cannot reach it
  • Timing: polling, stabilization, and cool-down
  • How a scale rule signs in
  • Jobs scale executions, not replicas
  • Exam-pattern recognition

The three scale rule categories in Container Apps

Decision axisHTTPTCPCustom
What it measuresConcurrent HTTP requests to the revisionConcurrent TCP connections to the revisionAny metric a ScaledObject-based KEDA scaler exposes, including CPU, memory, Service Bus, Event Hubs, Kafka and Redis
Threshold propertyconcurrentRequestsconcurrentConnectionsNamed by the scaler, such as messageCount for Service Bus or value for CPU
Default threshold1010Set by the scaler rather than by Container Apps
Can the app reach zero?YesYesEvent-source scalers yes; CPU and memory no
Polling interval appliesNoNoYes, every 30 seconds
Available to jobsNoNoYes, for ScaledJob-based scalers on an Event job

Decision tree

Which scale rule fits the workload?Each event needs its own container?starts, runs once, exitsIs the signal an event backlog?queue, topic or event sourceDoes it serve HTTP requests?or hold TCP connectionsCPU or memory ruletype=Utilization, floor of 1 replicaEvent-driven jobScaledJob scaler, min and max executionsCustom rule for that sourceService Bus, Event Hubs, Kafka, RedisHTTP rule, or TCP ruleconcurrentRequests or concurrentConnectionsyesyesyesnonono
Cheat sheet

Cheat sheet

  • minReplicas 0 plus an event scaler enables scale-to-zero
  • Only HTTP, TCP, and event rules can scale to zero
  • The azure-servicebus scaler scales on queue depth
  • Scaler auth uses managed identity or a secret
  • Topic-based scaling requires topicName plus subscriptionName
  • KEDA supports many event-source scalers
  • An HTTP scale rule scales on concurrent requests
  • Several rules on one app are OR-ed, never AND-ed
  • minReplicas and maxReplicas bound autoscaling
  • Scale-rule metadata sets the per-scaler threshold

Unlock with Premium — includes all practice exams and the complete study guide.

Also tested in

References

  1. Set scaling rules in Azure Container Apps
  2. Kubernetes Event-driven Autoscaling (KEDA) in Azure Kubernetes Service (AKS)
  3. KEDA: Scaling Deployments, StatefulSets & Custom Resources
  4. KEDA Azure Service Bus scaler
  5. Tutorial: Scale an Azure Container Apps application
  6. KEDA FAQ FAQ
  7. KEDA CPU scaler
  8. Billing in Azure Container Apps
  9. KEDA ScaledObject specification
  10. Managed identities in Azure Container Apps
  11. Jobs in Azure Container Apps