Compute
AWS App Runner
AWS App Runner deploys web applications from source code or container images with managed builds, TLS, and autoscaling. It offers a simple home for stateless GenAI APIs that do not require Kubernetes or detailed infrastructure control.
Key points
- Services expose managed HTTPS endpoints
- Automatic scaling follows request concurrency settings
- VPC connectors provide outbound access to private resources
When to use it
- Host a lightweight RAG web API
- Deploy an internal prompt-testing service from a container
Exam tips
- Prefer App Runner for straightforward web services; ECS or EKS offers deeper runtime and networking control
- Cold starts and downstream model quotas still need explicit latency and backpressure design