Moon: A Multi-Scenario Monitoring and Alerting Platform for Cloud-Native, IoT, and AI

Moon is an open-source monitoring and alerting platform in Go maintained by the aide-family team. It uses a monorepo architecture to integrate shared tool libraries and microservices, built on the Kratos framework. Key features include: supporting monitoring needs for three major scenarios (cloud-native, IoT, AI); balancing code sharing and independent service evolution in its architecture; providing bilingual documentation and a convenient development toolchain. Project address: https://github.com/aide-family/moon (Release date: 2026-05-27).

In modern distributed systems, monitoring requirements vary significantly across scenarios: cloud-native needs to focus on container/microservice health; IoT needs to handle massive device data collection; AI systems need to track model performance and inference latency. Existing solutions struggle to cover these scenarios uniformly, so Moon was born to address the integration problem of cross-domain monitoring.

Moon adopts a monorepo strategy, with advantages including easy code sharing, safe cross-service refactoring, and unified version management. Internal modular design: each microservice has its own go.mod, Makefile, and documentation, achieving "internal open source". Core components:

• Magic Box: Shared tool library (security, OAuth, etc.)
• Goddess: Authentication and authorization service
• Rabbit: Business service (time engine, etc.)
• Marksman: Event service
• Jade Tree: Collection and communication proxy runtime Component names are derived from Chinese mythology, making them easy to remember and culturally distinctive.

Moon's tech stack selection:

• Development language: Go 1.25+ (high concurrency, fast compilation)
• Microservice framework: Kratos (open-sourced by Bilibili, supports HTTP/gRPC dual protocols, unified configuration and logging)
• CLI tool: Cobra (standardized subcommand structure, easy for operation and maintenance management) Reasons for selection: mature community, fully production-verified, gentle learning curve, suitable for long-term maintained infrastructure projects.

Moon's cross-scenario capabilities:

• Cloud-native: Monitor Kubernetes clusters, container status, microservice call chains
• IoT: Handle massive device connection management and data collection
• AI: Track model performance metrics, monitor inference services Extensibility design: New monitoring types can be integrated via plugins/independent services without modifying core code; Marksman handles alert event flow uniformly, and Jade Tree provides flexible collection deployment options.

Development experience:

• Bilingual documentation (Chinese and English) lowers the barrier to participation
• Makefile commands: make all builds all applications, make <app> runs a single application, make gen generates code, reducing context switching costs Practical value: Provides a reference architecture for teams building monitoring systems; native support for AI scenarios (model latency, throughput, GPU utilization, etc.) meets the monitoring needs of large model inference services.

Moon is an open-source monitoring platform with clear architecture, practical tech stack selection, complete documentation, and wide scenario coverage. Whether used directly or as an architectural reference, it is worth in-depth study. Its monorepo multi-service organization, Kratos microservice implementation, and support for three major scenarios provide a valuable practical sample for modern monitoring platform design.