How ZTO Tech’s Quality Team Automated Dubbo Interface Testing with JMeter

1. Background

During interface testing of a distributed service framework Dubbo configured with full Spring, the team at ZTO (Zhongtong) faced the problem of “no direct API call – cannot test via URL + parameters like HTTP”.

2. Implementation Principle

The quality improvement department built an end‑to‑end automated and platformized solution that integrates Testlink, JMeter and Jira. The workflow is visual and WYSIWYG. The architecture is shown in the diagram.

3. Testing Process

3.1 Environment Management – supports local and global variable configuration; the Zookeeper URL for Dubbo is set here.

3.2 Interface Management – all seven Java request fields (address, interface name, method name, group, version, parameter types, request data) are configured in a single entry, enabling project‑driven script management.

3.3 JDBC Connection – provides database address, username and password for JMeter to query data.

3.4 BeanShell PostProcessor – extracts JSON data; scripts can be edited directly in the platform.

3.5 Assertion – BeanShell assertion compares the JSON response with the database query result.

3.6 Run Management – supports 24/7 unattended execution, custom schedule and environment selection.

3.7 Test Report – JMeter reports are extended with ANT integration to display task, case, environment and configuration details.

4. Conclusion

JMeter offers extensible components for custom interface testing. The ZTO quality center’s platformized solution has executed hundreds of Dubbo interface cases, uncovered dozens of defects, and proved its effectiveness. Ongoing work focuses on further UI and workflow optimization.