Interview with Liu Qi, Co‑founder and CEO of PingCAP, on TiDB’s Origin, Roadmap, and Database Trends

From December 22‑23, the GIAC Global Internet Architecture Conference was held in Shanghai, featuring experts from leading tech companies. In the lead‑up to the event, High Availability Architecture interviewed Liu Qi, co‑founder and CEO of PingCAP, about TiDB and the future of databases.

Liu Qi, co‑founder and CEO of PingCAP, is a noted Go and Redis expert who created TiDB (inspired by Google F1) and the open‑source projects TiDB and Codis.

He explained that TiDB originated from challenges with MySQL scaling at PeaPod, leading to the development of the distributed cache system Codis and later the desire to solve MySQL’s scalability by building a NewSQL system inspired by Google Spanner and F1.

Three years later TiDB 1.0 was released, and users have moved beyond simple MySQL sharding to use TiDB as an HTAP database for high‑concurrency transactional workloads and real‑time analytics, with notable deployments at companies such as Toutiao, Mobike, Meituan, and others.

Looking ahead, Liu outlined three main R&D directions: continuously strengthening TiDB’s SQL capabilities and real‑time analytical queries; adapting the storage engine to modern hardware and storage media; and deep integration with cloud platforms and Kubernetes to achieve a truly cloud‑native, self‑driving database.

He highlighted TiDB’s technical advantages: a share‑nothing architecture that provides elasticity and high availability; MySQL‑compatible protocol and syntax that lower migration cost; a custom distributed storage layer TiKV that outperforms traditional file systems; strong‑consistent distributed transactions; and the TiSpark project that enables SparkSQL analytics on the same data.

Successful use cases span internet and finance sectors, including large‑scale deployments at Mobike, Ele.me, Qunar, Tongcheng, and financial risk‑control systems at 360 Finance and others.

Regarding broader database trends, Liu sees a resurgence of SQL, the necessity of distributed scale‑out, ACID transaction support, cloud‑native design, and the convergence of analytical and transactional processing (HTAP) as defining the next generation of databases.

He also shared PingCAP’s open‑source governance: strict upstream project processes, branch management, code review, heavy reliance on GitHub, extensive automated testing with millions of test cases, and a talent pipeline that converts active contributors into full‑time engineers.

Finally, Liu previewed his GIAC talk on multi‑tenant technology in TiDB, emphasizing its importance for cloud databases, and expressed hopes for the conference’s continued success.