Full‑Link Load Testing Practices for iQIYI Payment System

iQIYI provides video, sports, live streaming, literature and other services to billions of users, generating massive and unpredictable traffic that can affect the stability of its payment system; accurate capacity assessment and full‑link pressure testing are therefore essential.

Full‑link pressure testing simulates peak‑traffic requests in a production‑like environment to evaluate system capacity and guide optimizations, but payment services face challenges such as complex traffic composition, mismatched traffic conversion, hidden bottlenecks in shared resources, and overall resource waste.

To address these issues, the team explored and practiced four key steps: (1) Core‑link mapping – identifying primary and branch service dependencies, mocking external channels, and handling side‑track services; (2) Test‑environment preparation – propagating a test identifier through APM, isolating data with shadow tables, and tagging messages; (3) Traffic construction – analyzing production logs to build realistic traffic models that reflect payment‑method distribution and business strategies; (4) Execution and protection – validating business and environment readiness, monitoring key metrics (service call volume, success rate, latency, message backlog, Redis and DB load), and ensuring fallback mechanisms.

During execution, the team performed business‑capacity verification and system‑limit testing, discovering bottlenecks such as order‑ID generation latency caused by Redis‑based sequences and RDB fork pauses; after iterative tuning, the payment system reached near‑design TPS limits.

Future work includes consolidating scattered requirements (data models, traffic scheduling, monitoring, performance analysis) into a unified, one‑stop solution to lower the entry barrier and cost of full‑link testing, as well as automating diverse payment‑request generation and providing detailed test reports for upstream services.