The request hit our desk at midnight: make this service visible to the right systems without ever exposing it to the wrong ones. No inbound ports. No public endpoints. No surface for the scanners to latch onto. Only outbound connections.
Discoverability with outbound-only connectivity sounds like a paradox. But it’s not. It’s the way to ensure systems can be located and interacted with by what matters—while staying off the radar of what doesn’t. It removes the attack vector of inbound traffic. No open sockets. No firewall holes. Only controlled, encrypted egress.
Traditional service discovery depends on inbound reachability. DNS names and IPs lead straight to something that listens. But listening means exposure. Outbound-only connectivity flips the trust model. The service initiates all connections out to a known registry, broker, or coordination plane. Systems learn about each other through shared, authenticated channels, not by scanning public networks.
With outbound-only discoverability, infrastructure changes less often fall apart under network reconfigurations. Firewalls, NAT, and changing IP addresses become trivial. Security rules become simpler because the direction is always out. The connection paths are predictable. There is always an intermediary that can mediate identity, authorization, and service metadata at scale.