ServiceDiscovery

Service discovery is a mechanism in distributed systems to enable components to locate one another's network locations. It addresses the dynamic nature of modern architectures where service instances may scale up or down, move, or fail. Central to service discovery are service registries, which maintain records for available service instances and their network addresses, metadata, and health status. Clients or routers query the registry to obtain current endpoints.

There are two broad patterns: client-side discovery and server-side discovery. In client-side discovery, a service instance queries the registry directly to obtain a list of healthy instances and selects one to call, often with client-side load balancing. In server-side discovery, a load balancer or reverse proxy queries the registry and routes requests to instances, shielding clients from endpoint details.

Mechanisms for implementation include DNS-based discovery, HTTP-based registries, and specialized systems. DNS service discovery uses DNS SRV or A/AAAA records to resolve service instances. Registry systems include Consul, Etcd, Zookeeper, and cloud-native options such as Kubernetes service discovery (via CoreDNS or kube-dns). Health checks and heartbeats keep the registry up to date.

Service discovery is often combined with service registration, health monitoring, and sometimes service mesh technologies that handle discovery, load balancing, and security policies across microservices.

Challenges include consistency, latency, and partition tolerance in dynamic environments, as well as securing service endpoints and preventing stale routing information.