GBase 8s High Availability Quartet – ER Enterprise Replication: The Connecting Link for High Availability

Published on 2026-05-21

GBase 8s High Availability Quartet | Part 4 (Finale)
ER — Enterprise Replication

The GBase 8s High Availability Quartet first three parts cover HAC for same-city protection, RHAC for remote-site extension, SSC for read scaling — together they form a high-availability moat within the cluster. But in the real world, data is often scattered across different database instances nationwide: the headquarters runs an SSC cluster, branches use standalone instances, and provincial front-end systems are deployed in government extranets — requiring synchronization of only partial tables, or even only certain rows and columns. Who builds the bridge? ER is the connecting link that ties all this together — the final piece of the high availability architecture.

Precision Delivery: Four Granularities, Transmit Only What’s Needed

ER is an asynchronous log-level replication engine built into GBase 8s: it directly reads logical logs, captures changes non-intrusively, and coordinates data across nodes on a per-transaction basis. Under normal loads, latency is kept within 1 second. After a network outage, data is automatically resynchronized without manual intervention — nearly maintenance-free.

Choose from four levels of granularity — database, table, row, or column — combined with row filtering and conditional indexes, to transmit only what’s needed with zero bandwidth waste. Three topology templates are available for flexible composition: Fully Connected Mesh (financial-grade strong synchronization), Hierarchical Tree (parent-child architecture for cross-region disaster recovery), and Forest Tree (multi-level tree fusion for enterprise-wide data aggregation). They cover every scenario from within a single data center to across continents.

Five-Dimensional Conflict Resolution: Multi-Master Writing Without Conflicts

Worried about conflicts in multi-master writes? ER includes a five-dimensional conflict resolution scheme:

For primary-standby setups, use "ignore conflicts" for unidirectional synchronization; for multi-point writes, use "timestamp + SPL" for intelligent arbitration. ER integrates seamlessly with HAC, RHAC, and SSC clusters; during cluster failover, ER automatically relocates to the new primary node — no manual intervention needed.

Four Business Models, Ready Out of the Box

  1. Data Distribution: Central publisher → Multi-node subscription, ideal for OLTP systems

  2. Data Consolidation: Multi-source aggregation → Central data warehouse, suited for business analysis

  3. Workflow Replication: Chain-based order passing, suitable for production system collaboration

  4. Everywhere Update: Global synchronization + conflict resolution, a true multi-master architecture

With that, all four parts of the GBase 8s High Availability Quartet come together

HAC guards the same-city baseline · RHAC extends the geographic radius

SSC handles read scaling · ER delivers flexible replication

Combine them as needed, building layer upon layer

From a single data center to a nationwide network, from intra-cluster to cross-instance

To erect a complete high-availability defense

Your path to business continuity is now fully paved