How Jiangsu Mobile Built an Integrated DevOps Monitoring Platform for Seamless Collaboration

In the era of agile development and integrated agile operations, Jiangsu Mobile explored and self‑developed an automated testing and detection framework based on DevOps principles, breaking the boundaries between development, testing, and production operations to achieve a full‑process closed‑loop management and a monitoring‑integrated platform that maximizes data sharing and reuse.

Core Component Technology Selection

Web Automation Testing Tool Selection

Using value engineering, mainstream open‑source web automation tools were compared on cost, ease of use, and extensibility, referencing industry cases. The Selenium open‑source tool was chosen for its quick recording, strong extensibility, cross‑browser support, and integration with JUnit, Feed4JUnit, and Java annotations. The framework was further customized to support DDT and keyword‑driven testing (KDT), separating test data from scripts to enhance collaboration and maximize asset reuse.

Interface Monitoring Tool Selection

Considering full coverage, security, extensibility, and strong business correlation, existing commercial and open‑source tools were insufficient. A self‑developed interface automation monitoring tool was created, offering unified case authoring and execution, dynamic tagging for modular organization, private protocol encapsulation, multi‑step orchestration, and regex‑based association to achieve data exchange and monitoring.

The platform also integrates the existing JMeter performance testing tool for unified scheduling, aggregating monitoring data from development, testing, and production environments for centralized performance diagnosis and root‑cause analysis.

Solution Overview

With large‑scale agile development, micro‑service transformation, and rapid business iteration, high‑quality continuous delivery becomes critical. The PAAS layer is divided into development, testing, and production domains. A keyword‑driven (KDT) automated testing framework was self‑developed to consolidate test assets and monitoring data across the three domains, forming a monitoring‑integrated platform that enables resource sharing, low‑cost, high‑quality, rapid application delivery.

Figure 1: Monitoring‑Integrated Platform Architecture

Development‑Testing Phase : Automated testing tools perform comprehensive regression testing. Developers and testers quickly create keywords; business analysts define complex workflows using these keywords, enhancing collaboration. Test cases are aligned with production performance metrics, enabling root‑cause location during testing.

Production‑Operations Phase : Test scripts are reused to continuously monitor production systems, allowing issues discovered in production to be reproduced in testing. Pre‑processed monitoring data feeds back to testing, improving early defect detection.

Closed‑Loop Management : The integrated platform enables bidirectional data flow. Production failures automatically generate test cases; performance metrics inform testing load models; faults are reproduced across domains for root‑cause analysis, achieving end‑to‑end closed‑loop management.

Figure 2: Closed‑Loop Management Process

Production Practice

Process Architecture

The platform adopts a distributed, decentralized, modular design consisting of three components: Agent (request simulation/data collection), Collector (model definition, execution scheduling, data transformation, workflow orchestration, case management), and Console (data visualization and system management).

Figure 3: Process Architecture Diagram

Automated Monitoring

Based on CRM business scenarios, web and interface services are categorized (personal, group, IoT, etc.). The focus is on automating monitoring for both web and interface services, significantly enhancing automation capabilities across development, testing, and operations.

Web Automated Monitoring

A self‑developed keyword‑driven testing architecture separates test steps, objects, actions, and data, maintained in Excel and unified into a test case set for reusability. Distributed agents simulate real user interactions across network segments, improving accuracy. Test results from both test and production environments are reported to a central server and displayed via a dashboard.

Figure 4: Web Automated Monitoring

Interface Business Monitoring

A unified interface test suite encapsulates environments, allowing flexible invocation across test cases and rapid switching between test and production via environment variables. Tags such as production center and regional dimensions differentiate execution and results. Visual dashboards display health status of top‑500 query interfaces across 13 regions.

Figure 5: Interface Test Suite

Figure 6: Interface Business Monitoring

The integrated monitoring platform bridges testing and operations, addressing rapid business iteration challenges while ensuring continuous functional verification. Automated tools and processes effectively connect development, testing, and operations, driving continuous product quality improvement.