This article explains the design and implementation of a timestamp‑based log replay engine that replays HTTP traffic using Java's DelayQueue, thread pools, and custom data structures, and provides performance test results demonstrating its ability to handle hundreds of thousands of requests per second.
The 2021 year‑end summary details FunTester’s rapid growth, the expansion of its public account, major performance upgrades reaching over 100 k QPS, the design of a distributed testing framework, dynamic load models, log‑replay innovations, and the author’s deep dive into Go, Java and Groovy.
This article details a Java‑based log replay solution that re‑implements Goreplay functionality using Disruptor, covering design concepts, log extraction, implementation steps, performance metrics, identified risks, and provides full source code examples for a scalable load‑testing engine.
The team built a log‑based replay framework that pulls and cleans old gateway logs, samples requests, concurrently replays them against both old and new gateways, automatically compares JSON responses with configurable ignore rules, retries failures, and uses matching response distributions to safely verify and migrate thousands of APIs.
The article describes how Ctrip's ticket front‑end team replaced a slow, manual log‑pulling process with a Flink‑based real‑time pipeline that streams Kafka data, indexes it in Elasticsearch, and enables second‑level log retrieval for automated scenario replay, dramatically reducing CI cycle time.
The article details CynosDB’s cloud‑native PostgreSQL design—using compute‑storage separation, log sinking with asynchronous replay, and multi‑version reads—to enable a one‑primary multi‑read architecture that delivers elastic scaling, reduced I/O, stateless instances, and rapid failover and efficient resource utilization through parallel log‑based recovery.
