Service Routing, Health, and Scaling

Configure paths, alternate routes, health checks, capacity, autoscaling, and network exposure.

Service runtime settings are environment-specific. Review each environment before deployment, especially when production uses different routing or capacity from development.

Routing

For public web services, configure the application port, domain, and path that should receive traffic. Alternate routes can direct additional host or path patterns to the same service. Keep routes unambiguous and avoid overlapping patterns unless precedence is intentional.

Static targets may also expose controls for the root file, not-found file, routing mode, and file-extension behaviour. Validate deep links directly rather than testing only the home page.

Health Checks

Set a health path that returns success without requiring a signed-in user. The endpoint should verify the process can serve requests, but it should remain fast and avoid expensive dependency checks.

Return one of the configured success codes.
Do not redirect to a login page.
Keep readiness and deep dependency diagnostics separate when possible.
Test the exact path through the deployed route.

Capacity and Autoscaling

Depending on the service target, configure desired, minimum, and maximum capacity together with CPU and memory. Autoscaling can use a CPU or memory threshold within the configured bounds.

Set the minimum high enough to meet availability requirements and the maximum high enough for expected bursts without exceeding downstream or budget limits. Scaling a service does not automatically scale its database, queue consumers, or other dependencies.

Network Exposure

Public access, outbound internet access, allowed IPs, and internal-only routing are separate decisions. Enable only what the workload requires. Apply AWS WAF to public HTTP workloads that need managed or custom filtering.

Validate a Change

After deploying infrastructure, deploy a release and confirm:

the public and internal routes resolve as expected;
the health check is passing;
the service has reached desired capacity;
logs and metrics are arriving;
autoscaling bounds match the intended environment.