Ensuring Online Education Service Quality through Real Traffic Capture and Replay

Online education services require high quality and efficient development processes; the article begins by emphasizing the importance of CI/CD maturity and the need to balance quality and efficiency using real traffic data.

Two main traffic capture methods are discussed: main‑line replication (log reporting, code instrumentation, framework‑level duplication) and side‑car replication (network protocol capture). The side‑car approach using Nginx access logs is chosen for its low intrusion and language‑agnostic nature.

The article outlines the core capabilities needed for traffic‑based quality assurance, including user‑level and scenario‑level traffic collection, sequence identification, and OpenAPI integration.

Business model construction is detailed, showing how core fields are identified, data is labeled, and traffic is recolored and offset to create replay‑ready scenarios while preserving real‑world characteristics.

Traffic replay execution relies on processed traffic data, with thread control to balance link chaining and execution efficiency, and includes mechanisms for response validation, JSON schema generation, and alerting on verification failures.

A diff service compares replay responses against historical baselines, handling noise, multiple comparison modes, and large‑scale result extraction, providing quantified anomaly indices.

The platform, named Conan, offers these capabilities as services, employing a distributed backend architecture with master‑worker nodes, load balancing, and Kubernetes‑based scaling to ensure stability and high availability.

Additional platform features include pressure modeling based on real traffic, performance testing integration, and various application scenarios such as continuous testing, performance test linkage, and regular business assurance.

The article concludes with a call for broader collaboration and plans to open‑source the platform to foster community involvement.