How JD Cloud Engineered a Seamless 618 Shopping Surge: Ops Strategies & Disaster Drills

Background

During the COVID‑19 pandemic, delivery workers became essential, and JD.com’s 618 shopping festival required a “smooth blood flow” of services to support massive consumer demand.

Resource Preparation

Early procurement and reuse of server hardware were prioritized. Existing equipment was relocated, data erased, OS installed, and delivered to business units. New machines were ordered, and despite pandemic‑induced uncertainties, delivery timelines were met, following a “reuse first, purchase less” principle.

Hardware Fault Management

The fault pool grew as thousands of devices failed weekly. Restrictions in data‑center access increased difficulty. JD classified faults, reserved spare parts, assigned dedicated contacts, and expedited repair procedures. Within two months, over ten thousand faults were resolved, keeping the fault pool below a safe threshold.

Network Engineering and CDN Scaling

Anticipating the 618 peak and concurrent political meetings, JD coordinated with carriers to reserve bandwidth and performed early network expansion. Detailed, module‑by‑module expansion plans allowed hour‑level scheduling and parallel cut‑overs before carrier lockdowns. Core, access, and aggregation devices were inspected, hardened, and monitored 24/7.

CDN, the backbone for static content, was expanded several‑fold across hundreds of data centers nationwide. Traffic models guided bandwidth upgrades, and extensive stress testing ensured reliable “last‑mile” delivery during the peak.

Technical Drills and Disaster Recovery

Extensive capacity‑testing and failover drills were conducted. Scenarios included automatic removal of faulty cluster nodes, rapid traffic redistribution, and simulated core‑data‑center loss. The system detected and isolated failures within seconds, and a four‑hour exercise uncovered hidden risks, confirming high availability for cloud‑based services.

Monitoring and big‑data analytics were employed to achieve minute‑level fault detection, analysis, and remediation, further strengthening operational resilience.

Organizational Coordination

A “1+1+N” guarantee organization was formed: one decision‑making group, one coordination group, and multiple departmental guarantee teams. Cross‑departmental emergency plans were created, covering core‑link traffic monitoring, high‑risk mitigation, and alert handling, with continuous drills to identify and improve weak points.

Outcome

The combined hardware, network, CDN, and disaster‑recovery preparations enabled JD Cloud to support the 618 event without major incidents, demonstrating a mature, data‑driven operational framework for large‑scale e‑commerce traffic spikes.