Get the latest tech news

400TB Single Cluster: OceanBase Powers Kwai`s Core Business

Kwai is a short video app boasting more than 10 million daily active users. How does it efficiently process highly concurrent user requests? Kwai once deployed multiple MySQL clusters in the backend to support high traffic with large data storage and satisfactory performance. What are the weak points of this conventional sharding solution? What pushed Kwai to select distributed databases and eventually deploy OceanBase Database? In this post, Xiaochong, the head of Kwai's O&M team, shared the team’s reflection and experience in implementing the OceanBase Database solution.

In this case, we had to squeeze many MySQL nodes into a single cluster, and we could not ensure the timely return of business requests during peak hours, not to mention that heavy-duty hardware was required for the middleware or downstream links to operate stably. To ensure system stability and reliability, we needed to connect the MySQL database to ClickHouse, Elasticsearch, or Doris, which might require more data replicas, and thus further increased the hardware costs. We realized that the sharding solution was more like a painkiller rather than a cure, and we were in urgent need of a high-performance, scalable distributed database system that not only meets our business requirements but is also easy to operate and maintain.

Get the Android app