An emergency-alert paraphraser API is bursty and benefits from scale-to-zero between incidents. Comparing App Service Linux containers and Container Apps, what is the better default for that pattern?

For a bursty emergency-alert paraphraser that should scale to zero between incidents, Azure Container Apps is the better default versus classic App Service Always On with permanent minimum instances. Container Apps can idle at zero replicas when no alerts fire, then scale out on demand, aligning cost with sporadic municipal emergency traffic. That matches the stem's primary requirement: scale-to-zero between bursts.

Preferring classic App Service Always On with minimum instances forever fails by concept for a scale-to-zero goal: Always On and permanent minima keep capacity billed and warm even when no alerts are firing. Storing paraphrases only in Excel on a file share with no container host fails because it is not a container hosting comparison and cannot serve a bursty API. Shutting down Azure entirely until the next declared emergency fails because it is an organizational outage, not an elastic hosting pattern, and rebuild latency would hurt alert response.

Exam caveat: App Service can still host containers well when you need Always On, slots, and web-app features; the distinguishing requirement here is scale-to-zero. Cold start after zero may add latency—acceptable for many emergency paraphrase bursts if warm pools are not mandated. Validate regional quotas and scale limits before relying on rapid scale-out during real incidents.

Operational check: Deploy the paraphraser on Container Apps with a scale rule that allows zero replicas, simulate idle then a burst of alert payloads, measure scale-out time and success rate, and compare monthly idle cost against an Always On App Service baseline. Confirm runbooks cover cold-start expectations for emergency ops.