// the find
alibaba/SREWorks
Cloud Native DataOps & AIOps Platform | 云原生数智运维平台
SREWorks is Alibaba's internal SRE platform, open-sourced after nearly a decade of internal use on their Big Data infrastructure. It combines application lifecycle management on Kubernetes (via OAM) with DataOps pipelines and AIOps capabilities — think PagerDuty + Flink + a k8s app manager, all bundled together. This is for large engineering orgs that want a self-hosted ops platform and have the k8s chops to run it.
The OAM-based app manager is the genuinely interesting piece — it gives you a typed abstraction over k8s workloads that goes beyond raw Helm/kubectl, with built-in support for blue-green and canary deploys. The Flink integration for real-time anomaly detection is battle-tested at Alibaba scale, not a toy demo. Helm chart decomposition is well-structured — each subsystem (appmanager, kafka, nginx) is its own sub-chart, so you can disable what you don't need. It ships with actual data pipeline tooling rather than just dashboards pointing at Prometheus.
The documentation is almost entirely in Chinese — the English README points to a yuque.com doc site that is Chinese-first, which is a real adoption blocker for teams outside China. The dependency footprint is enormous: Kafka, Flink, MySQL, Redis, Elasticsearch, and a custom k8s operator all required before you see a UI. Activity has slowed significantly — commits are sparse in 2024-2025, and the last push in December 2025 doesn't reflect sustained development. The 'AIOps' layer is vague in the open-source version; the truly intelligent bits likely live in Alibaba Cloud services that aren't included.