Data Middle Platform Construction Project of Chengdu Rural Commercial Bank

In its established "1-3-5" strategic development goals, Chengdu Rural Commercial Bank emphasized the steadfast implementation of a digital strategy for three rural sectors (agriculture, rural areas, and farmers) finance, enabling financial services to build confidence in the development of a digital countryside. However, as the bank's business grew rapidly and diversified, its legacy systems could no longer meet evolving operational demands. Capitalizing on this opportunity, the bank launched a data middle platform project. As a core component of this platform, the data warehouse needed to efficiently and accurately integrate and process massive volumes of bank-wide data, delivering timely data support for applications and serving as the heartbeat of the entire platform's operation. To achieve this, the data warehouse had to run on a high-performance, stable computing engine. Through research into advanced industry peers, combined with the experience of leading vendors, an MPP database was identified as the optimal choice for the bank's data warehouse computing engine.

• Boost Database Performance: Support massive data volumes and complex analytical processing with high-performance parallel computing.

• Adopt Columnar Storage: Increase query efficiency and reduce storage costs through data compression.

• Boost Data Loading Efficiency: Enable timely ingestion of massive data.

• Provide Standardized Interfaces and Broad Compatibility: Seamlessly integrate with mainstream middleware, ETL, and reporting tools.

Chengdu Rural Commercial Bank deployed GBase 8a MPP Cluster, a distributed parallel database from General Data Technology, as the data warehouse for its data middle platform. The project went live in 2022, enabling accurate consolidation and processing of massive bank-wide data to deliver timely, reliable support for data applications. Built on a Shared-Nothing architecture with columnar storage and high-performance compression, the solution provides high-throughput, high-concurrency data extraction, loading, and querying—keeping maximum response times under three seconds and ensuring a satisfying user experience. In this architecture, the GBase 8a MPP Cluster consists of four computing nodes. The application service layer serves user queries through query interfaces, while a data distribution server shards and dispatches user requests to multiple nodes for parallel execution.

• High Satisfaction: Empowers data-to-business enablement, dramatically boosting customer satisfaction. 

• Performance Boost: Adopts MPP databases with high-performance parallel computing, drastically reducing system query response times.

• High Scalability: Supports linear scalability; cluster performance improves linearly as nodes are added.

• Cost-Effective: Uses open PC server hardware platforms, with costs far lower than minicomputers with traditional databases, and performance far exceeding legacy architectures.