Case Study: Zhihu boosts big data performance with PingCAP’s TiDB

Zhihu cuts large query execution time from 30-40 seconds to 6-7 seconds with PingCAP

Zhihu, China's largest knowledge sharing platform, faced performance and stability challenges within its big data infrastructure. Their Apache Hive and HDFS systems were strained by large, resource-intensive queries, which slowed down task execution and risked cluster instability. To address this, they turned to vendor PingCAP and its distributed SQL database, TiDB.

PingCAP implemented TiDB as a high-performance metadata store for their systems. This enabled Zhihu to build a real-time alert system for large Hive queries and to cache HDFS content summaries, drastically reducing pressure on the NameNode. As a result, the execution time of large queries decreased from 30-40 seconds to 6-7 seconds, and the latency for certain HDFS RPC requests dropped from several minutes to under 10 milliseconds.

Zhihu