// the find
didi/KnowStreaming
一站式云原生实时流数据平台，通过0侵入、插件化构建企业级Kafka服务，极大降低操作、存储和管理实时流数据门槛
KnowStreaming is a Kafka management platform from Didi, covering multi-cluster operations, monitoring, ACLs, consumer group management, and partition rebalancing through a web GUI. It targets platform/SRE teams running Kafka at scale who want operational visibility without writing CLI scripts. The project is Chinese-first — documentation, community, and most of the codebase comments are in Chinese.
Broad Kafka version support (0.10.x through 3.x, both ZK and KRaft modes) without requiring changes to the Kafka deployment itself. The plugin architecture means enterprise features (load rebalancing, HA mirroring) are separate modules, so you're not blocked from using the open-source core while they upsell. The metric collection pipeline separates collection from sink (ES), which is a reasonable design choice that makes it easy to swap storage backends. It ships with a live demo environment, which is rare and genuinely useful for evaluating before committing to a deployment.
Elasticsearch is a hard dependency for metrics storage — there's no option to use something lighter like VictoriaMetrics or Prometheus, which is a significant operational burden for teams that don't already run ES. The enterprise feature annotations (EnterpriseLicense, EnterpriseLoadReBalance) scattered through the codebase signal that meaningful functionality is behind a commercial license, but the boundary isn't documented clearly anywhere. The last meaningful commit activity has dropped off, and the community channels are WeChat/Zhishi Xingqiu — essentially inaccessible to non-Chinese speakers, making it hard to get help or gauge project health. The setup requires MySQL (state), Elasticsearch (metrics), and Kafka itself — three external dependencies before you've done anything useful.