How ICBC Built a Unified Multi‑Database Test Data Management Platform

Background and Challenges

In recent years, the Industrial and Commercial Bank of China (ICBC) Software Development Center has been transitioning its information systems from a centralized architecture to an open‑platform distributed architecture, gradually adopting domestic database products. During this transformation, maintaining test data in development and testing environments faced several major challenges:

Increasing business scenario complexity : Business verification requires coordination across multiple product lines, demanding flexible upstream and downstream data maintenance and a multi‑role, multi‑person data maintenance channel.

Multiple database client switches : Coexistence of various database products leads to a proliferation of client tools with differing operations, greatly increasing usage and maintenance difficulty.

Database user password storage risk : Numerous development and testing environments and product lines require storing multiple sets of sensitive information (usernames, passwords, connection strings), raising the risk of leakage.

Data tampering risk : Frequent data modifications in test scenarios and many project members can cause unintended data changes that affect other test cases.

There was an urgent need for an efficient collaboration mechanism suitable for multiple database types, multiple business product lines, and large‑scale development‑testing teams.

Practice: Building the Test Data Maintenance Platform

ICBC's Software Development Center explored solutions to support flexible data modification while ensuring security and control.

1. Establish a unified online database maintenance platform : The platform centrally manages both traditional and domestic databases, breaking client tool silos, and consolidates database inventories, user passwords, and other sensitive information. After authentication, users can log in to any product database with a zero‑learning‑curve experience, dramatically reducing data maintenance costs for developers and testers.

2. Implement tiered management of critical data : Important tables are marked with protection identifiers; modifications require test manager approval. All data changes are managed and executed online, providing traceability and rollback capability while maintaining a smooth modification experience.

3. Offer diversified multi‑channel data services : The platform can expose specific data‑change statements as services for other project teams to perform self‑service data maintenance, schedule automated execution, and provide standardized database connection services for various automated testing tools, ensuring compatibility across testing channels.

Results

After 11 months online, the test‑environment data maintenance service has achieved visual management of hundreds of product lines and nearly ten thousand database nodes, serving thousands of users with over one million data queries and one hundred thousand data modifications. The solution delivers convenient data maintenance while guaranteeing security and control, solving the challenge of efficiently maintaining test data for large‑scale development teams across multiple database types and business lines, and contributes to cost reduction, efficiency gains, and digital transformation for the enterprise.