G-Tech Moment | How Does GBase 8s Achieve 99.999% High Availability Architecture?

Published on 2026-02-27

In critical sectors such as finance, telecommunications, and energy, which are fundamental to the national economy and people's livelihood, business systems must ensure 24/7 continuous service, requiring the underlying database to achieve 99.999% availability. Against the backdrop of a growing shift towards adopting China-developed databases, these databases must match or surpass their international counterparts in high availability technology.The GBase 8s database has built a full spectrum of high availability capabilities, delivering a multi-layered and customizable 'non-stop' assurance solution for the smooth migration and secure operations of core business systems across industries.

The First Layer: Primary-Standby Clusters for Flexible Local and Intra-city Disaster Recovery

GBase 8s offers primary-standby clusters based on real-time log replication, where the core is data synchronization between primary and standby nodes via redo logs. It supports two modes to fit different scenarios: HAC for same-city high availability and RHAC for remote disaster recovery.

HAC Cluster: It provides three data synchronization modes, allowing you to choose based on the trade-off between data consistency and performance:

  • Full Synchronization: Before a transaction commits, the log must be replayed on the standby node, achieving RPO=0 (zero data loss). Ideal for scenarios requiring the highest data consistency.

  • Near Synchronization: Before a transaction commits, the standby node only needs to acknowledge receipt of the log, ensuring very low risk of data loss while delivering faster response times.

  • Asynchronous: Transactions are asynchronously replicated to the standby after commit on the primary, providing optimal performance. Suitable for disaster recovery scenarios where slight data delay is acceptable.

  • Features: Simple deployment, with failover typically completed within 10 seconds. It also supports read-write separation, effectively offloading query workloads.

RHAC Cluster: A single-primary, multi-standby asynchronous cluster designed specifically for remote disaster recovery. It does not require real-time acknowledgment, thus having minimal impact on primary node performance. It easily handles common latency and bandwidth limitations of long-distance network transmissions, ensuring eventual data consistency across sites.

The Second Layer: Shared Storage Cluster (SSC) to Power High-Performance Core Operations

For core businesses with massive data volumes (typically 10TB to 100TB) and demanding performance and scalability requirements, GBase 8s offers the Shared Storage Cluster (SSC) based on shared storage. This architecture is comparable to Oracle RAC, with multiple nodes sharing the same set of stored data.

  • All nodes can read and write, delivering excellent linear scalability.

  • It uses a master-peer management model; when a single node fails, other nodes quickly take over (RTO <5s), ensuring business continuity.

  • Only a single copy of the data is stored, ensuring strong consistency while eliminating data redundancy. It is a highly cost-effective cluster solution for high-performance core business systems.

The Third Layer: Real-Time Data Exchange and Sharing for Active-Active and Flexible Data Flow

In addition to high availability at the cluster level, GBase 8s also features built-in real-time data synchronization at the table granularity, comparable to OGG (Oracle GoldenGate). This capability is highly flexible:

  • Supports one-way or bidirectional synchronization between multiple nodes, enabling true active-active database architecture.

  • Can synchronize full rows or selected columns, supporting full loads, incremental syncs, and checkpoint restart.

  • Failover time is extremely short (within 1 second), maximizing business continuity.

  • It is ideal for real-time data exchange between headquarters and branch offices, data sharing between parallel entities, or scenarios requiring minimal manual intervention such as remote unattended sites.

The Fourth Layer: A Combined Approach – Building the Ultimate Two-Site, Three-Center High Availability Solution

For systems with the strictest SLA requirements, GBase 8s can combine the above technologies to build a comprehensive two-site three-center solution.

A typical deployment architecture is: a 2-node SSC cluster in the primary data center to provide high-performance services; an HAC node in the same-city disaster recovery site to achieve near-synchronous or fully-synchronous protection; and an RHAC node in the remote disaster recovery site to protect against regional risks. All access is directed through a unified Connection Manager (CM), transparent to applications. Any single node or data center failure can be automatically switched within the designed time, achieving the highest level of availability for business operations.

From flexible primary-standby clusters, to high-performance shared storage clusters, to granular data synchronization and multi-site disaster recovery, GBase 8s provides a comprehensive and flexibly configurable high availability product stack. This not only means that when replacing Oracle databases, users have a functionally equivalent, or in some scenarios, a more advantageous technology choice, but also signifies that China-developed databases now have the solid capability to support the 'always-on' operation of the nation's critical core business systems.