Technical Analysis of the Xi'an Health Code System Crash and Its Performance Bottlenecks

Recently the Xi'an "One‑Code" health platform suffered a second major outage, prompting the author—who identifies himself as a senior architect—to investigate the technical reasons behind the failure.

Initial speculation on social media suggested that the backend was transmitting images directly, but deeper analysis revealed that the real issue was the delivery of static resources (JavaScript, CSS, images) from a single, non‑CDN endpoint.

By estimating the total size of these assets at roughly 500 KB and the peak request volume at about 33 000 queries per second, the author calculated the required bandwidth as 33 000 × 500 KB × 8 bits ≈ 125 Gbps, a load far beyond the capacity of a single data‑center network card, causing the interface to become saturated and the service to go down.

The discussion references several Zhihu answers that confirm the QR‑code generation endpoint /person/app/refreshQRCode is not the culprit; instead, the lack of CDN for static files is the primary bottleneck.

Images from the original post illustrate the error messages and the network‑traffic analysis, emphasizing that while the problem appears simple, it reflects a fundamental architectural oversight in handling high‑traffic public services.

The author concludes that without moving static assets to a CDN and properly scaling the network edge, similar large‑scale systems will continue to experience catastrophic failures under peak load.