Rebuilt kernel strengthens the foundation, an independent ecosystem opens new horizons—exploring the advanced path of GBase 8s database

On August 21, the 16th China Database Technology Conference (DTCC 2025), themed “Intelligent Innovation, Data-Driven Future,” was held in Beijing. More than 100 industry technology leaders delivered “2+16” keynote sessions, drawing 2,000+ database technology experts and practitioners from across China.

The conference featured a dedicated GBase Session, presenting “GBase 8s: From Kernel Rebuild to Ecosystem Independence” across application practice, product evolution, technical breakthroughs, high availability, and ecosystem development.

Proven in Production: GBase 8s Powering Shenzhen Metro’s AFC System

The Shenzhen Metro team presented Chinese IT Application Innovation Database Practice in the Shenzhen Metro AFC System, sharing how GBase 8s serves as the core data platform across multiple mission-critical systems, including AFC, CLC, and ACC.

GBase 8s fully meets Shenzhen Metro’s requirements for functionality, performance, reliability, and compatibility. At the station level, high availability is delivered through same-city replication (HAC); at the center level, a multi-tier HA architecture combines shared storage clustering with same-city replication (SSC+HAC). With the one-stop migration tool (MTK) and the GEM monitoring and operations platform, the system supports ticketing, fare gates, station entry/exit, operations management, passenger flow analytics, and revenue analysis for tens of millions of urban metro passengers. This deployment breaks the long-standing dominance of foreign mainstream databases in China’s rail transit systems, delivering lower cost and stronger performance with end-to-end localization from backend to frontend—demonstrating GBase 8s’ reliability in critical infrastructure such as rail transit.

A Decade of Focus: GBase 8s’ Path of Independent Innovation

Cui Zhiwei, General Manager of the GBase 8s Product Division, delivered the keynote GBase 8s Over the Past Decade, outlining the evolution and core capabilities of this Chinese centralized database.

According to Cui, the decade of GBase 8s mirrors the journey of China’s database industry—from technical exploration to mature production adoption. Since the first release in 2014, GBase 8s has iterated from V8.3 through V8.8, advancing independently across compatibility, storage engines, the optimizer, drivers and interfaces, peripheral tools, ORM middleware, and the developer ecosystem. It has grown into a secure, reliable, and production-proven centralized transactional database.

GBase 8s has delivered substantial Oracle and MySQL compatibility enhancements, and made strong progress in primary-standby and shared storage clustering—building highly compatible, highly reliable, and cost-efficient cluster solutions with enterprise-grade strong consistency. In performance stress tests, it achieved RPO=0 and RTO < 3s. A robust tooling ecosystem enables full-lifecycle graphical management for data migration and sync, data development, and database monitoring and operations. Today, GBase 8s is deeply deployed in core systems across finance, rail transit, energy, and government, becoming a trusted choice for enterprise customers.

Engine Rebuild: Technology Innovation for Production Workloads

Yang Wei, Head of GBase 8s R&D, presented Outcomes and Lessons from the GBase 8s Storage Engine Rebuild, detailing technical advances of the new storage engine for production environments.

“The storage engine is the foundation of a database—it defines the capability ceiling over long life cycles,” Yang said. To meet the extreme demands of core production systems for high concurrency, low latency, and high reliability, the GBase 8s team completed a two-year storage engine rebuild.

The new engine combines an in-memory design with a traditional disk-based engine and delivers in-memory MVCC. It rebuilds the primary-standby failover mechanism to achieve 3-second failover under stress-test load; adopts lock-free programming and dynamic resource allocation to scale high concurrency to the 10,000+ level; resolves consistent reads on standby nodes, so users no longer struggle with read/write split deployments; expands support for broader online DDL to protect business continuity; and enables ultra-fast flashback reads without waiting for lengthy log rollback.

Key work in the rebuild includes snapshot reads under mixed read/write workloads, multi-version management, appropriate GC strategies and timing, and hybrid storage engines on standby nodes. The storage engine team continues to deepen the product in areas such as high-concurrency indexing, HTAP under multi-NUMA, row-column hybrid processing, and transaction continuity.

From Available to Reliable: GBase 8s High Availability Clusters Keep Business Online

Huo Zenghui, Product Director at GBase, delivered From Available to Reliable: GBase 8s High Availability Cluster Solutions, detailing three core HA offerings.

Shared Storage Cluster (SSC) targets core transaction scenarios that require strong consistency with data volumes from 2TB to 100TB. It supports global transaction processing, multi-node read/write, and automatic failover. In stress tests: RPO=0, RTO < 3s, and TPS of 3,000+ in banking core scenarios. The SSC-based database appliance is ready out of the box for extreme performance needs.

Primary-Standby Replication Cluster (HAC) targets same-city disaster recovery for datasets under 2TB. It supports three sync modes, with DML redirection on the standby; automatic failover (RPO=0, RTO < 8s); and simplified operations. The HAC-based cloud platform solution helps customers move to the cloud with strong cost efficiency.

The real-time data exchange and sharing solution is powered by GBase 8s’ built-in asynchronous replication engine (ER). ER supports multiple collaboration modes and delivers low-latency sync at database, table, row, and column granularity.

Today, GBase 8s HA cluster solutions are deeply embedded in core systems across finance, energy, transportation, retail, and other critical industries, safeguarding continuous business operations with proven reliability.

Collaborative Ecosystem: Breaking Silos to Build a New Partnership Model

Guo Zhuo, Director of the GBase 8s Product Division, presented Database Ecosystems: From Technology Silos to Collaborative Symbiosis, outlining GBase 8s’ ecosystem strategy and practice.

Guo noted that the essence of a database ecosystem is “connection”: rooted in the database core, it brings together developers, upstream and downstream partners, interfaces, frameworks, and tools into a coordinated whole—maximizing