An architecture style where the system is a fleet of small, independently-deployable services communicating over the network.
Microservices is an architectural style where a system is decomposed into small, independently-deployable services, each owning a specific business capability and communicating with the others over network APIs (REST, gRPC, async events). The pattern emerged from organizations (Netflix, Amazon, Spotify) that hit the limits of large monoliths and needed independent team velocity, fault isolation, and per-service scaling. The tradeoff is operational complexity: every microservice is a deploy target, an observability source, and a potential incident.
Microservices unlock scale (in both teams and traffic) but tax the operations side, the network becomes the bottleneck, latency budgets fragment, distributed tracing becomes mandatory, and incident response now spans multiple owners. Most teams are right to start with a modular monolith and split into microservices only when team scale forces the issue. Once you do split, invest in service-mesh observability and graceful-degradation patterns from day one.
See the part of the platform that handles microservices in production.