How a Team Built End‑to‑End Load Testing from Scratch at a Fast‑Growing E‑Commerce Platform
This article details how a performance testing team at a major e‑commerce company designed and evolved full‑link load testing from zero to a mature, repeatable process, covering its definition, historical milestones, challenges during major sales events, and future optimization plans.
What Is Full‑Link Load Testing?
Full‑link load testing simulates massive user requests and data against the entire production‑like environment to uncover bottlenecks, verify scalability, and ensure system stability across the whole business chain. It reduces costs, improves service availability, strengthens team collaboration, and ultimately enhances user experience and business value.
From Zero to One: Building the Capability
Before June 2019, the company had never performed full‑link testing. As traffic surged during major sales (e.g., 618, Double 11), the performance team was formed in April 2019, set up standards and environments by May, and completed the first large‑scale test in June. The first official full‑link test for Double 11 occurred in October 2019.
Leveraging that experience, the team executed four full‑link tests in 2020 (including the “Wucai Stone” project, 618, the five‑year anniversary, and Double 11), all of which proceeded smoothly.
Key Challenges Faced
During the 2019 Double 11 event, four major challenges emerged:
Complex core‑link mapping due to intricate business scenarios and high service dependencies.
High cost of replicating production environments (1:1 mirrors), data preparation, and service integration.
Difficult traffic estimation caused by missing tracing tools, insufficient monitoring coverage, and lack of experienced personnel.
Multi‑track execution involving service and database sharding, Redis large‑key management, and version‑merge releases.
Addressing these required thorough core‑link analysis, request‑level monitoring, and upstream/downstream dependency analysis to assess traffic.
Full‑Link Testing Process
The end‑to‑end testing workflow is divided into four stages: initiation, preparation, execution, and readiness. Each stage emphasizes safety thresholds such as maximum processing capacity, stable processing capacity, and horizontal scalability.
Evolution Timeline
The testing practice evolved as follows:
2019 Double 11 – first full‑link test.
2020 “Wucai Stone” architecture refactor – practical experience gained.
2020 618 – smooth execution based on prior lessons.
2021 five‑year anniversary – rapid three‑day turnaround from notification to production testing.
2022 Double 11 – completed full‑link testing within 15 days.
This progression illustrates a shift from initial exploration to a repeatable, scalable process.
Future Plans
Upcoming improvements focus on:
Technical optimization: multi‑protocol support, traffic recording and replay, standardized SOPs.
Scenario coverage: broader breadth, deeper depth, and combinatorial scenarios.
Data preparation: automated embedding, cache warm‑up, and synthetic data generation.
Process efficiency: standardization, automation, and routine integration.
Continued refinement is expected to further enhance testing reliability and business confidence.
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
DeWu Technology
A platform for sharing and discussing tech knowledge, guiding you toward the cloud of technology.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
