Service registry

Intent

Keep a directory of service instances so clients/routers can discover where to send requests.

Use when

• Service instances are dynamic (autoscaling, containers, ephemeral IPs) and you can’t hardcode locations.
• You need a single place to coordinate discovery, health, and routing decisions.

Avoid / watch-outs

• The registry becomes critical infrastructure; plan HA, failure modes, and operational ownership.
• Discovery alone doesn’t solve reliability; clients still need timeouts and sane retry behavior.

Skill mapping

• architecture: choose discovery approach (client-side vs server-side vs mesh).
• platform: provide client/router helpers so discovery is consistent across services.
• resilience: timeouts + circuit breakers around calls to discovered instances.
• observability: metrics/logs for lookup failures, routing decisions, and health state.