How tencentblueking/bkmonitor-datalink works
Project type and purpose
• This is not a monolithic full-stack app but a suite of Go-based microservices and SDKs that together form a “data link” pipeline for the BlueKing observability platform. It provides:
• Data collection (metrics & events)
• Third-party alert ingestion
• Data transformation and routing
• High-availability proxy to InfluxDB
• A unified PromQL-style query service
• Task scheduling and control via a Kubernetes operator
File structure (pkg/ directory)
• Top-level service modules:
• bkmonitorbeat – host-side collector agent (Beats-style)
• ingester – HTTP/Kafka pull or push of alert events
• transfer – core ETL: convert data formats and publish to storage back ends
• influxdb-proxy – HA proxy & load-balancer for InfluxDB clusters
• unify-query – HTTP/gRPC query service providing PromQL/STQL over InfluxDB
• bk-monitor-worker – job scheduler & worker pool for periodic or event tasks
• bkmonitor-operator – Kubernetes Operator (CRDs, controllers, Helm)
• libgse & kits – shared SDKs and helper libraries
• Each module is a standalone Go microservice (cmd/ or main.go), uses dependencies (Redis, Kafka, Consul, HTTP, gRPC), and ships its own Dockerfile/Helm chart.
README for architecture notes
• Design goals: plug-and-play modules, decoupled via config, MQ, standard protocols