Data Platform Replacement Project for a Commercial Bank in Fujian

Project Background

Established in 1996 with the approval of the People's Bank of China and the local government, the bank is a city-level joint-stock commercial bank in Fujian Province. It was renamed a bank in Fujian Province in 2009 with the approval of the China Banking Regulatory Commission. The bank now operates a headquarters, six branches, and 41 sub-branches, covering all districts of the city. It adopts a flat operational structure consisting of head office, branches, and sub-branches.

Key Challenges

The bank’s existing data warehouse, which runs on Oracle, suffered from inconsistent data logic across downstream applications, slow performance on large-scale queries, long batch processing windows, and limited visualization of critical business metrics. To ensure regulatory reporting quality, meet internal analytics needs, integrate and improve data models, and optimize critical batch processing paths, the bank required a new big data platform. The goal was also to enable business metric analysis, large-scale wide-table queries, and mobile distribution of key operational indicators to address increasingly complex business demands.

Solution Requirements

Based on the bank’s actual situation, the new big data platform had to meet the following objectives:

High Performance: Significantly improved processing capacity compared with the existing data warehouse.

High Scalability: The platform must be highly scalable to allow seamless expansion according to evolving business needs.

High Availability: The platform should run on domestically manufactured servers, ensuring both strong availability and compliance with the bank’s domestic IT application innovation requirements.

Service Assurance: Local, manufacturer-direct technical support capable of promptly resolving all issues encountered during operation.

The project's big data platform is built with GBase 8a MPP Cluster, initially deployed on 11 nodes, and uses virtual clusters to partition resources to meet the needs of various business systems. Leveraging GBase 8a MPP's parallel processing, columnar storage, intelligent indexing, and high-efficiency compression, it accelerates batch job processing and improves query performance while supporting ad-hoc queries and multidimensional analysis to power complex BI analytics and visualization.

Architecture diagram of the big data platform for a bank in Fujian Province

Deployment Overview

The big data platform was launched in 2023 with 11 nodes deployed, totaling approximately 90 TB of data.

Key Results and Value

Significant Performance Improvement: Leveraging GBase 8a MPP Cluster's capabilities in massively parallel processing, columnar storage, smart indexing, and high-efficiency compression, it meets the customer's data processing and query performance requirements, improving overall efficiency by more than 5 times compared to the original data warehouse.

Reduced Costs: Long-term customer validation has proven that GBase 8a MPP Cluster runs stably on cost-effective domestic servers, significantly reducing hardware investment costs and meeting the bank's requirements for China's IT Application Innovation (信创) standards.

Dynamic Scalability: GBase 8a MPP Cluster supports horizontal scaling of database nodes without service interruption, ensuring business continuity.

High Availability: The multi-replica redundancy policy of GBase 8a MPP Cluster ensures database service continuity even in the event of node failures.