GBase 8s Centralized Transactional Database: Two-Site Three-Center Solution

Published on 2023-05-30

Requirements

Database systems are the core components of information systems and a prerequisite for normal system operations. Therefore, high security and reliability are required for database systems, especially in core business systems. A centralized database places all data in a single central location, with all business operations connecting to this database for read and write operations. This model offers the advantage of simple management and maintenance, but it introduces risks of single points of failure and performance bottlenecks, failing to ensure system sustainability. To address these issues, enterprises typically deploy centralized databases using a two-site three-center architecture to ensure business continuity.

Solution Overview

The primary site deploys an SSC cluster for load balancing, with multiple nodes in the cluster providing read and write services. The same-city disaster recovery site hosts HAC nodes for hot standby, ensuring rapid switchover if the primary site fails. Typically, the standby site nodes are read-only, serving data query, reporting, and other functions, thus offloading pressure from the primary site. The remote disaster recovery site deploys RHAC nodes, enabling long-distance disaster recovery over thousands of kilometers. It supports forced manual switchover to become the primary site or automatic switchover in case of primary site equipment failure. All sites are centrally managed by the CM Connection Manager, which routes service requests to specific cluster nodes based on SLA rules.

Key Features

  • Hot Standby Support

The highly available database cluster features hot standby. In the event of a system failure, it can quickly and automatically switch to the standby node, restoring database services to ensure business continuity and minimize customer impact from interruptions.

  • Data Consistency Between Active and Standby Systems

When a customer completes a transaction, data is synchronized in real time across all nodes. If a database switchover occurs due to a failure, the customer can continue subsequent services seamlessly without data inconsistency, eliminating the need to restart the transaction.

  • Automatic Switchover

The primary-standby switchover in the cluster is automatically handled by the database management system without manual intervention. The switchover is transparent to the application system, requiring no modifications. For example, if business operations are interrupted due to database hardware or software failure, once the database recovers, the application automatically resumes normal operation.

  • Sub-Second Switchover Time

Ensures real-time data synchronization between standby and active nodes, maintaining data consistency. Supports fault detection and automatic takeover by the standby node. RTO < 20s, RPO = 0.

  • Long-Distance Disaster Recovery

When the production data center loses all database service capabilities due to power outages, equipment damage, or other reasons, the high-availability database cluster can provide disaster recovery and switchover over hundreds or even thousands of kilometers, ensuring business continuity.