// the find
apache/dubbo-go
Go Implementation For Apache Dubbo .
dubbo-go is the Go port of Apache Dubbo, a Java-origin RPC and service governance framework widely used in Chinese internet companies. It gives Go services first-class participation in a Dubbo ecosystem — service discovery, load balancing, traffic management — and interops with Java Dubbo via the Triple protocol. Target audience is teams running polyglot microservice stacks where Java Dubbo is already the backbone.
The Triple protocol is the right bet: it's gRPC-compatible (protobuf, HTTP/2) but also HTTP-friendly, so you can curl a service directly without a gRPC client. Service discovery supports Nacos, Zookeeper, Etcd, and Polaris out of the box — no vendor lock-in at the registry layer. The HA cluster strategies (failover, failback, forking, broadcast) are all there and individually tested. Configuration hot-reload with content-based caching to suppress spurious notifications is a real operational nicety that most frameworks skip.
The documentation is overwhelmingly Chinese-first — the English README points to `cn.dubbo.apache.org` for the Go SDK docs, which is a friction wall for non-Chinese teams. The JavaScript-based script routing (Goja engine) is a surprising dependency for a Go microservices framework and adds a non-trivial attack surface if routing rules come from user input. HTTP/3 is explicitly experimental in 3.3.1 and shouldn't be trusted in production yet. If you're not already running a Dubbo Java ecosystem, the setup cost is steep: you need Nacos or Zookeeper just to get service discovery working, which is a lot of infrastructure for teams used to Kubernetes-native service meshes doing that for them.