How AI Is Transforming Databases: Highlights from China’s DTCC2025

On August 23, 2025, the 16th China Database Technology Conference (DTCC2025) concluded in Beijing, offering a three‑day showcase of forward‑looking topics and deep‑dive practice sessions for domestic database professionals.

01 AI Leads the Database Wave

The conference highlighted a clear trend: AI is comprehensively reshaping database technology paradigms, moving from an auxiliary tool to a core driver of database design and operations.

Tencent Cloud presented TDSQL‑C, a serverless, full‑link architecture that independently senses compute, memory, and storage resources, achieving second‑level scaling for million‑level concurrency and millisecond‑level latency scenarios.

Huawei Cloud demonstrated GaussDB’s latest AI‑integrated and vector database capabilities, embedding vector retrieval into the core engine to unify scalar and semantic queries.

02 Vector Databases and RAG Become Focal Points

With large‑model technology penetrating industry scenarios, Retrieval‑Augmented Generation (RAG) is rapidly becoming a core architecture for building intelligent applications. A dedicated “Vector Database and RAG Retrieval Technology Practice” session was one of the most popular.

In sectors such as finance, power, and government, challenges like data integration, high system reuse cost, and complex business value assessment arise. Huawei GaussDB’s “intrinsic fusion” concept redefines the capability boundaries of enterprise‑level knowledge retrieval.

Huawei Cloud’s database intelligent operation experts shared a case where a DBA assistant boosted operation efficiency by 300%, enabling AI‑driven anomaly inspection and risky SQL detection to pre‑warn database health risks.

03 Distributed Database Performance Practices

As the host of the “Outstanding Distributed Database Performance Practice” session, I witnessed strong industry interest, including AI‑related shares and lively Q&A.

AutoHome’s Tao Huixiang presented how TiDB replaced traditional architectures to handle terabyte‑scale data and high concurrency, supporting forums, HTAP real‑time analytics, and major events, achieving performance gains, cost reduction, and high availability.

Tencent Cloud’s kernel R&D expert Wang Dianceng discussed TDSQL‑MySQL’s evolution from version 2.0 with weak distributed query optimization to version 2.5 with a revamped query architecture that markedly improves performance.

China Mobile’s Fu Hui‑jun described the internally developed Panwei distributed database (based on OpenGauss) that replaced Oracle for core billing systems, delivering 2%‑55% performance uplift, supporting 20 billion daily call records, and achieving 99.999% availability.

Kongzhong Cloud’s Zhao Feixiang compared Spanner, YugabyteDB, and OceanBase, outlining core concepts, architecture models, consistency protocols, and offering selection methods, optimization strategies, and future trends such as cloud‑native, autonomous, and multimodal data processing.

Memobase founder Ye Jianbai introduced the storage demands of long‑term AI agent memory, advocating for multimodal, multi‑structure data management beyond traditional RAG, with PostgreSQL extensions and the emerging Memobase+EloqDB solution addressing latency, cost, and structural integration.

04 The Road to Database Autonomy

In the “Database Autonomy” session, Nanda General shared a case on GBase 8s enabling continuous transactions and zero data loss for financial core systems, leveraging four core engine values to build a comprehensive, multi‑layered financial‑grade database solution.

Huawei’s financial database solution director emphasized the importance of mastering key factors in database solution architecture selection to achieve sustainable development.

05 Intelligent Operations and Emerging AI Agents

AI is fundamentally changing database operation modes. Huawei Cloud’s scenario‑driven operation agents have increased DBA fault‑locating and recovery efficiency by 300%.

The agents support AI‑driven anomaly inspection, risky SQL identification, and proactive health risk warnings, with an intelligent Q&A accuracy of 90%.

06 Interview on GoldenDB and OpenGauss

Within the “Intelligent Innovation, Numbers Win the Future” theme, GoldenDB’s vice‑general manager Tu Yaofeng discussed the AI‑driven core engine, domestic substitution practice, and future trends of AI‑database convergence.

An interview with OpenGauss R&D director Dr. Xiong Qin covered open‑source community competitiveness, the breakthrough of DataVec vector capabilities in OpenGauss 7.0, the high‑availability and disaster‑recovery benefits of the oGRAC multi‑write architecture, and the roadmap for database evolution.

07 Summary

Reflecting on the conference, databases have evolved from passive data carriers to intelligent agents that actively optimize business. Vendors are driving the shift toward AI‑driven database partners, and the Chinese database ecosystem is thriving in both technology and talent development.