AZ-305 Practice Question 18: Event Driven Container Platform
Microsoft Azure Solutions Architect Expert (AZ-305) · Test 2 · question 8 of 10
A ticketing company is replatforming eight containerized microservices. Each one reads from an Azure Service Bus queue, and the queues are empty for most of the day but receive heavy bursts when an event goes on sale. The architecture review requires that:
- instances scale out on queue length and scale in to zero when the queues are empty, so nothing is billed between bursts, - services discover and call each other and support gradual traffic splitting between revisions, - the team does not manage, patch, or upgrade cluster nodes, - no direct access to Kubernetes APIs is needed.
Which compute service should you recommend?
Show answer and explanation
Answer:
Container Apps is the serverless container platform for exactly this profile. It is built on Kubernetes with KEDA, Dapr, and Envoy underneath, and it enables event-driven architectures by scaling on traffic and on event sources such as queues, including scale to zero. It also supplies the application-level features the review asks for (service discovery and traffic splitting between revisions) as a fully managed experience, and it deliberately does not expose the Kubernetes APIs, which the team does not need.
Option A delivers the same scaling behavior but not the same operational model. With AKS the cluster lives in the subscription and the team owns node pools, upgrades, and configuration; the node pool also cannot bill nothing at idle unless it is scaled to zero nodes, which adds cold-start and scheduling complexity the platform otherwise handles.
Option C cannot reach zero. An App Service plan bills for its allocated instances whether or not requests arrive, and App Service autoscale is oriented around web traffic and resource metrics rather than queue depth. App Service is optimized for web applications, not for queue-driven background microservices.
Option D is a building block, not a platform. Container Instances provides individual container groups on demand, with no built-in scale, load balancing, revisions, or certificates, so the option has to invent the orchestration in a Function, which is the operational burden the team was trying to avoid.
Reference: Comparing Container Apps with other Azure container options