JD.com 618 Shopping Festival Technical Preparation: Resource Optimization, High Availability, and System Reliability

In 2020, JD.com held its 17th "618" shopping festival, marking the first major sales event in the post‑pandemic era and setting a new record of 269.2 billion yuan in cumulative order value, creating an unprecedented traffic surge.

Despite not purchasing any new compute servers, the JD R&D system endured the pressure through three months of preparation, contributing to the platform’s explosive business growth.

The festival’s success highlighted the importance of resource fine‑grained management and high‑availability governance. Since 2019, JD has not added hardware; instead, it reclaimed nearly 10,000 servers via the “Gold‑Mine Plan,” improving container core counts by 25% and reducing IT resource costs by 19.5% year‑over‑year.

Technical highlights included a newly formed Architecture Committee that reviews cross‑departmental readiness, incorporates low‑CPU‑utilization metrics into assessments, and promotes a culture of cost awareness.

High‑fidelity load‑testing played a crucial role: the team built a full‑chain testing platform, conducted five end‑to‑end drills, covered over 20 core business links, and increased testing tasks by 51% compared with the 2019 event, ensuring front‑ and middle‑office systems could handle peak loads such as the Maotai product rush.

Front‑end and middle‑office systems were tightly coordinated, pre‑emptively scaling resources and applying rate‑limiting to maintain stability during the 1.99‑times traffic spike relative to the 2019 event.

Looking forward, JD plans to launch the “Taishan Project,” a long‑term initiative aimed at further improving system reliability through modular deployment, automated high‑fidelity testing, platform‑wide disaster‑recovery, and standardized failover mechanisms.