GBase 8s Multi-Read/Write Solution

With the explosive growth of data in recent years, storing and processing massive datasets has become a major challenge across industries. This puts higher demands on traditional centralized transactional databases, particularly for heavy read and write workloads, which manifest in the following ways:
 

1. As businesses expand, databases must support larger-scale read/write operations and higher concurrency—requirements that a single database often cannot meet.

2. For mission-critical systems, continuously available database services are essential. Single-point failures can have significant impact, making standby databases and failover mechanisms a necessity.

3. Data statistics and reporting scenarios generate a high volume of read requests, which can lead to resource contention. A read/write separation model is needed to boost database performance.

GBase 8s Enterprise Replication (ER) is a log-based asynchronous tool for replicating data between GBase 8s database servers. On the source server, Enterprise Replication reads the logical log, stores transactions, and reliably transmits each transaction as replication data to the target server. This flexible, streamlined architecture enables

message broadcast from one source server to multiple target servers, or message consolidation from multiple source servers to a single target server.

Both broadcast and consolidation modes fall under unidirectional data replication. The key advantage is simpler management with no data conflicts.

Unidirectional Data Replication Model

Message Broadcast

Message Consolidation

In addition, we can support bidirectional data replication across multiple servers to achieve real-time data synchronization across all nodes—an any-point update model that creates a peer-to-peer application environment. Compared to unidirectional replication, this model handles more complex business systems and supports multi-read, multi-write workloads. However, it also involves more complex management and configuration, and data conflict considerations. To address conflicts, the replication system provides a rich set of pre-built conflict detection and resolution strategies, which customers can easily configure according to their actual business needs.

Any-Point Update Model

Fully Connected Structure

Hierarchical Tree Structure

Forest Tree Structure

• Asynchronous Data Replication

Data replication across nodes automatically occurs within the ER network without waiting for commits from each other.

• Log-Based Data Capture

Incurs minimal additional resource overhead and places no extra load on the source server.

• High Performance

Leverages a scalable, parallel-processing architecture to handle multiple data replication streams concurrently.

• High Availability

Asynchronous replication tolerates network and target node failures. Data is either stored on the target or rolled back to the source’s send queue; once normal operation resumes, changed data is automatically pushed to the remote node.

• Consistent Data Distribution

Source node data can be routed to the target through multiple intermediate nodes in the ER network, guaranteeing data consistency throughout distribution.

• Repair and Initial Data Synchronization

GR supports full data synchronization, especially when deploying GR into a live production environment with ongoing DML operations; you can use GR’s full sync to establish initial consistency.

• Flexible Architecture

The replication topology supports a forest topology, enabling data replication from a leaf node in one tree to a leaf node in another, traversing multiple NOROOT nodes along the way.

• Centralized Management

Administrators can manage and monitor all nodes in the replication system via a Windows NT graphical user interface (GUI) or a Unix command-line interface (CLI).

• Easy Implementation

Multi-node unidirectional or bidirectional replication clusters can be easily built using the GBase 8s graphical configuration and management tool.

• Network Encryption

Supports SSL encrypted transmission.