Service Governance in Microservices: Registration, Load Balancing, Rate Limiting

Monolithic Architecture

Before adopting microservices, many projects use a simple SpringBoot + SSM stack: a single JAR is built, deployed, and started with java -jar. SpringBoot embeds Tomcat, which listens on a port and handles HTTP requests.

Microservice Architecture

Microservices split the monolith into many independent services. Each service registers itself with a service registry (Nacos) and communicates via RPC (Dubbo). Historically, Dubbo + Zookeeper was common; recent projects favor Spring Cloud Alibaba, which combines Nacos for registration and Dubbo for RPC.

Service Governance – Registration, Discovery & Load Balancing

The first step of service governance is registration and discovery. Services register their address in Nacos; other services discover them via Nacos and invoke them through Dubbo.

Load balancing is the second step. When a service has multiple instances, Dubbo selects one instance using a load‑balancing algorithm before making the RPC call.

Service Governance – Rate Limiting & Circuit Breaking

Rate limiting protects the system from overload. For example, if the system can handle 1,000 requests per second but receives 2,000, excess requests are rejected.

Circuit breaking prevents cascading failures. If a downstream service repeatedly fails, the circuit opens and subsequent calls fail fast, avoiding further strain on the caller.

Both rate limiting and circuit breaking are provided by Sentinel in the Spring Cloud Alibaba stack.

Configuration Center

Traditional projects store configuration in src/main/resources files, requiring a rebuild and redeploy to change settings. Introducing a configuration center (Nacos) allows dynamic loading of core configuration at runtime, so updates can be made directly in Nacos without repackaging.

Service Monitoring

Monitoring covers server metrics (CPU, memory, disk, network, I/O), JVM GC, and application‑level metrics such as QPS and latency. It also includes tracing of inter‑service calls.

Prometheus collects resource and JVM metrics, while SkyWalking provides distributed tracing of microservice call chains.

By combining Nacos, Dubbo, Sentinel, Prometheus, and SkyWalking, developers can achieve a complete service governance solution for microservice systems.