How JD Cloud Powers the 11.11 Mega Sale: Scaling, High Availability, and Monitoring Strategies

Another 11.11 shopping festival is approaching, bringing tight timelines and heavy tasks for the promotion preparation. This article explains how JD's Zhilian Cloud, the technical backbone of the event, quickly gets teams into shape, identifies system risk points, and readies resources.

First, the classic eight‑step preparation guide is revisited: from resource assessment and simulated pressure testing to plan organization, online rehearsal, and final post‑promotion review.

Identifying the protection scope means defining the boundaries of the promotion, distinguishing core business lines, and classifying level‑0 and level‑1 systems to prioritize protection work.

For example, estimating a traffic demand of 10 Gbps or 1 M QPS allows the team to evaluate required bandwidth, instance count, cache size, and database capacity, prompting early scaling or procurement to ensure resources are ready before the promotion.

Core plans focus on high availability: automatic primary‑standby switching in cloud databases (RDS) and manual one‑click failover when needed. Services are deployed across multiple regions (Beijing, Guangzhou, Suqian, Shanghai) and Availability Zones (AZs) to mitigate single‑point failures.

Business degradation is another crucial safeguard: when traffic exceeds capacity, limiting or rejecting a portion of requests preserves service for the majority of users, and connection‑pool limits protect downstream databases.

Elastic scaling plans ensure that any unexpected traffic spikes during the promotion can be addressed by rapid online scaling, though most capacity is provisioned in advance.

Monitoring serves as the system’s eyes. Survival monitoring checks IP, port, and service health, as well as AZ status. Performance monitoring tracks metrics such as TP99 latency, CPU usage, memory consumption, and disk I/O, while business‑specific monitors detect abnormal task states.

After resource assessment, monitoring, and plan preparation, a full‑chain pressure test simulates the estimated order volume, observes metric changes, and validates alarm and remediation procedures. System‑specific rehearsals, such as RDS failover and data recovery drills, ensure automatic switching and backup restoration work flawlessly.

Post‑promotion, the team conducts a comprehensive review to identify improvement areas for future events.

The 2022 challenges included massive resource demand, prompting a shift to cloud services that treat infrastructure as a commodity, enabling cost‑controlled, on‑demand provisioning. Automated DevOps tools quickly map business dependencies, allowing rapid identification of risky level‑0/1 systems.

Early “open‑door” rehearsals (6.1 and 11.1) provided realistic load tests, revealing issues that were promptly fixed and confirming the system’s readiness for the main 11.11 peak.

Finally, the adoption of collaborative tools (mobile, email, online documents) improved remote communication and progress tracking, reinforcing the need for continuous inheritance and iterative optimization of the promotion preparation process.