Size a service registry: heartbeat rate, watch notification fan-out and registry memory footprint.
A registry's steady-state load is liveness traffic — every instance checks in once per interval — while its bursty load is watch fan-out, where one instance change notifies every watcher of that service. Memory scales linearly with instance count. Registries like Consul and etcd are consensus-backed, so every heartbeat is a replicated write; halving the heartbeat interval doubles the write rate against a quorum that has a hard ceiling.
Service Discovery
heartbeat rate = registered instances ÷ heartbeat interval, where instances = services × instances per service.
Watch fan-out
watch notifications = churn per second × watchers per service
heartbeat rate = registered instances ÷ heartbeat interval, where instances = services × instances per service. A registry's steady-state load is liveness traffic — every instance checks in once per interval — while its bursty load is watch fan-out, where one instance change notifies every watcher of that service. Memory scales linearly with instance count.
Registries like Consul and etcd are consensus-backed, so every heartbeat is a replicated write; halving the heartbeat interval doubles the write rate against a quorum that has a hard ceiling.
This calculator takes 6 inputs: Registered services, Instances per service, Heartbeat interval, Instance churn per minute, Watchers per service, Registry bytes per instance. The pre-filled defaults are a realistic starting point — replace them with figures from your own environment for a result you can act on.