Bank of China Head Office's Independently Developed Big Data Platform

According to the overall plan for Bank of China's enterprise architecture construction, by building a "three horizontals, two verticals, one line" digital asset operation and service system, it lays a solid foundation for the bank's data governance work.

Bank of China Head Office's Indigenous Big Data Platform

1. Project Overview

To align with Bank of China's enterprise architecture strategy, a digital asset operations and service framework was established to solidify the foundation for bank-wide data governance.

The framework is built around a unified group data platform with a three-tier architecture — “Data + Analytics + Presentation” — providing comprehensive, agile, and granular capabilities for data asset sharing, analytical applications, service delivery, and value creation. The data warehouse forms a critical part of the data layer.

The project deployed servers powered by domestic chips (Hygon), a domestic operating system, and a domestic database, delivering a fully homegrown enterprise data warehouse platform.

2.2 Solution

Bank of China Enterprise Data Warehouse System Architecture Diagram

Bank of China Enterprise Data Warehouse System Architecture

The GBase 8a cluster serves as the core of Bank of China's enterprise data warehouse, responsible for structured data storage management and computation. It supports enterprise-wide public data storage, integrated model computing, and shared data model architectures, while also directly handling query and analysis for certain business applications.

The Hadoop platform is used for data lake storage and the public data exchange zone. Data from various business systems is aggregated via TBDS distributed storage, and after initial processing, it is loaded into the GBase 8a cluster. The analytical results from GBase 8a are then stored back into TBDS distributed storage for consumption by other systems.

2.3 Results

  • Implementation

The project currently deploys over 200 servers powered by Hygon chips (China-made) and runs a domestic operating system. The GBase 8a v9.5.3 cluster consists of more than ten coordinator nodes, over 100 data nodes, and nearly 100 free node servers. To accommodate the domestic server architecture, each data node runs two instances.

The data warehouse holds approximately 100TB of historical data with daily increments exceeding 10TB. Over 6,000 batch jobs have been deployed, including around 2,000 core jobs, all completing within a 3‑hour window.

  • Benefits and Value

Massive Scale Cluster: The ultra-large-scale cluster delivers massive data storage and powerful computing capabilities, with horizontal scalability to over a thousand nodes, supporting Bank of China's future growth plans;

Fully Homegrown: A pioneering large-scale deployment in the financial industry combining China-developed servers, operating system, and database;

Lakehouse Integration: Seamlessly connecting the data warehouse and data lake to achieve unified lake and warehouse fusion.