Azure Container Apps (ACA)
What it is
Azure Container Apps is a managed, serverless platform for running containerized applications. You deploy an image and application configuration; Azure manages the underlying platform rather than requiring you to create or operate a Kubernetes cluster.
Why it exists
Many APIs, background processors, event-driven services, and microservices need more application lifecycle support than a standalone container group but do not need direct Kubernetes control. Container Apps provides managed ingress, revisions, jobs, and scaling for those workloads.
Key ideas
- Container Apps supports HTTP and TCP ingress, internal-only endpoints, and DNS-based service discovery within an environment. Configure network exposure deliberately; an app does not become private just because its data store is private.
- Apps can scale from HTTP traffic, events, CPU or memory load, and supported KEDA scalers. Most apps can scale to zero, but CPU- or memory-based scaling cannot. Assess current pricing and minimum-replica settings for the chosen design.
- Revisions preserve versions of a container app. A team can split traffic between revisions for blue/green deployment or A/B testing instead of replacing every user’s version at once.
- Jobs run on demand, on a schedule, or in response to events. They suit bounded work such as a nightly import; a long-running API is an app, not automatically a job.
- Dapr is an optional microservices integration. It can help with patterns such as service invocation and pub/sub, but it does not remove the need to design message handling, data consistency, and authorization.
- Container Apps can pull from public or private registries, including Azure Container Registry, and can use a virtual network when the environment is created with one.
Northwind example
Northwind runs its product-catalogue API in Container Apps. It exposes HTTPS ingress, deploys a new revision, and sends 10% of traffic to that revision while monitoring errors. A separate scheduled Container Apps job refreshes product data overnight. The API can scale from HTTP traffic, while the team keeps a minimum replica only when their latency and availability requirements justify it. The catalogue image is pulled from ACR using an authorized identity.
Container Apps versus AKS and ACI
| Need | Start by considering |
|---|---|
| Managed container app with revisions, ingress, jobs, and event-driven scaling | Azure Container Apps |
| Direct Kubernetes platform control and Kubernetes-specific configuration | [[aks |
| One simple container group without broader application lifecycle features | [[aci |
| Small event-driven code function without packaging a container yourself | [[azure-functions |
Common confusion
- “Serverless” does not guarantee zero idle cost or instant startup. Scaling behavior, minimum replicas, triggers, and pricing all matter.
- A revision is an application version, not a backup of a database or a complete rollback plan for data changes.
- A traffic split helps reduce release blast radius; it does not test all customer paths automatically.
Related
aci · aks · azure-functions · app-service · acr · app-service-plan
Sources and re-audit
- Microsoft Learn — Azure Container Apps overview, checked 2026-09-16.