GBASE supports the remote active-active project for the Rural Credit Banks Funds Clearing data platform

The Rural Credit Banks Funds Clearing Center Co., Ltd. (hereinafter referred to as the “Center”) provides clearing services for remittances, bank drafts, and universal deposit and withdrawal of personal accounts for 30 rural cooperative financial institutions and tens of thousands of branches nationwide. Guided by the national “Three Rural” policy—supporting agriculture, rural areas, and farmers—which continues to drive rapid rural economic growth, the Center has experienced a surge in both transaction volume and data volume.

The data platform is becoming increasingly critical to the Center’s business development. With growing demands for data security and remote disaster recovery, there is an urgent need to build a geo-redundant disaster recovery system for the data platform to provide comprehensive protection for data assets.

The Center plans to deploy a geo-redundant active-active system across Beijing and Changsha, balancing the dual requirements of business development and data security in the next stage. For the data platform, a multi-layered, comprehensive high-availability and data security solution is required, specifically:

• Cluster high availability — node failures do not impact business operations;

• Remote disaster recovery support with incremental data synchronization;

• Active-active deployment across locations, where both primary and standby clusters can provide external database services, and cluster switchover is supported;

• During active-active data synchronization across sites, concurrency control and synchronization interruption are supported while the primary cluster maintains continuous service availability.

The core data management layer of the Rural Credit Banks Funds Clearing Center Data Platform is built on the GBase 8a MPP Cluster, a massively parallel distributed database. Serving as the foundational data platform, it consolidates data from multiple core business systems, including the payment and settlement system, electronic commercial draft platform, online banking platform, online interbank payment clearing platform, self-service financial service platform, and shared multi-media financial service platform. It enables message-level granular data management and performs data integration and common processing for upper-layer applications, creating a unified foundational data platform for all enterprise data.

Data Platform Architecture for Rural Credit Banks Funds Clearing Center

On top of the foundational data platform, both fixed-reporting and ad-hoc reporting platforms are built to enable multi-dimensional analysis of business information and fast record-level querying of historical transaction data.

The Rural Credit Banks Funds Clearing Center Data Platform deploys primary and standby active-active clusters based on GBase 8a MPP Cluster in Beijing and Changsha, with the 8-node Beijing cluster serving as the primary.

The primary cluster at the Beijing center handles applications and batch data operations (including data loading and processing). After processing is complete, it incrementally synchronizes data to the standby cluster in Changsha at designated times each day. The dedicated line bandwidth between Beijing and Changsha is 20-30 Mbps, with a daily synchronization window from 12:00 noon to 24:00.

The GBase 8a MPP Cluster's active-active remote deployment uses a master-slave (M-S) synchronization architecture. The primary cluster supports read and write operations, while the standby cluster provides query access. The standby cluster can also offload external query workloads from the primary, reducing resource contention between reads and writes. Data synchronization is powered by a proprietary inter-cluster synchronization tool, which delivers core synchronization capabilities and exception handling with low coupling, enabling flexible API calls and supporting both intra-city and remote disaster recovery.

• Table-level incremental data synchronization: The synchronization tool enables users to define custom sync tasks and prioritize critical tables for replication.

• Atomicity: If a synchronization failure occurs for a table, the tool automatically rolls back the operation, ensuring data consistency between the primary and standby clusters.

• Incremental data capture: Changed data is automatically identified and synchronized.

• High-efficiency synchronization: Only primary copy data is replicated; replica copies are synchronized internally within the standby cluster. Synchronization uses compressed binary files directly, significantly improving throughput.

• Data validation: A checksum verification is performed upon completion to ensure the correctness of synchronized data between clusters.

The Rural Credit Banks Data Platform was built to address the diverse, time-sensitive, and complex data analysis demands of the bank's business departments and member institutions, enabling high-level applications such as real-time capital flow analysis, risk management, credit rating, and Internet finance. The remote disaster recovery system has been put into operation, synchronizing approximately 200 tables and over 100 GB of compressed data files daily. This achieves the following results:

Comprehensive high availability: The platform remains operational in emergencies—disasters, data center outages, network or power failures, and server node crashes. The active-active GBase 8a remote cluster between Beijing and Changsha delivers cluster-level high availability, while GBase 8a’s built-in node-level and process-level mechanisms ensure continuous service.

Enhanced data security: Remote disaster recovery backup further strengthens data protection.

Efficient synchronization: By synchronizing only primary replicas of critical tables, using point-to-point sync, incremental data sync, and compressed-file transfer, the primary-standby cluster synchronization is completed even with just 100 Mbps bandwidth.