OceanBase HTAP: Technical Insights Behind the TPC‑H World Record

On May 20, 2021, the Transaction Processing Performance Council (TPC) announced that Ant Group’s self‑developed distributed relational database OceanBase set a new world record in the TPC‑H analytical benchmark with a performance score of 15.26 million QphH.

OceanBase is also the only Chinese‑developed database to have achieved the top ranking in both transaction processing and data analysis benchmark tests.

The core member of the OceanBase testing team, Chen Mengmeng, shares the technical thinking behind this achievement.

OceanBase aims to be an HTAP (Hybrid Transaction and Analytical Processing) database, supporting both OLTP and OLAP workloads in a single system, reducing ETL complexity and enabling real‑time analytics on fresh data.

HTAP was first introduced by Gartner in 2014, highlighting its advantages in avoiding costly ETL and providing faster insights, which are becoming core competitive advantages for enterprises.

The article reviews the historical development of relational databases, from Codd’s relational model and IBM’s System R to Jim Gray’s transaction processing theory and the emergence of data warehouses, illustrating the long‑standing separation and later convergence of TP and AP workloads.

As hardware capabilities (SSD, large memory, many‑core CPUs) improved, the need for databases that can handle both TP and AP in a distributed fashion grew, leading to the rise of HTAP systems.

OceanBase, started in 2010 within Alibaba/Ant Group, evolved from an internal project to a commercial product, achieving record‑breaking performance in both TPC‑C (transaction processing) and TPC‑H (analytical) benchmarks.

Key technical advances include a cost‑based query optimizer with dozens of logical rewrites, a vectorized execution engine with cache‑aware optimizations, and resource isolation between TP and AP workloads via virtualized resource groups.

The article also discusses three major challenges for HTAP systems: massive data capacity and scalability, efficient execution engines for large‑scale analytical queries, and balancing the differing characteristics of TP (high‑throughput, low‑latency) and AP (complex, long‑running queries).

Future directions for OceanBase focus on strengthening HTAP capabilities while preserving its strong TP performance, simplifying usage for customers of all sizes, and continuing to innovate in areas such as distributed storage (OceanBase File System) and horizontal/vertical scalability.

Overall, the piece provides a comprehensive technical narrative on why HTAP is the future of databases and how OceanBase is positioning itself within this evolving landscape.