JDV Visual Big‑Screen Platform: Architecture, Challenges, and Technical Innovations for JD.com’s 618 Promotion

JDV (Visual Big‑Screen) is an internal JD.com platform for building data‑driven visual dashboards, offering more than ten template effects, over forty chart and navigation components, and supporting drag‑and‑drop, data switching, linked refresh, and drill‑down to meet the visualization needs of executives, sales, and product teams across the group.

The platform must satisfy business scenarios such as second‑level data updates, cross‑midnight count stops, rapid data adjustments, high data stability, and instant backup‑screen switching, all of which demand a flexible, high‑performance system.

During the 618 promotion, JDV faced major challenges: building and maintaining over 80 screens within a month, handling diverse and rapidly changing use cases (command‑center, celebration, PR), and coordinating communication among many teams, which created high communication overhead.

Key technical innovations were introduced: 1. Request state control – detecting inactive page states (hidden, minimized, etc.) to pause unnecessary polling; 2. Real‑time screen instance monitoring – a socket‑based heartbeat mechanism reporting client ID, version, IP, screen ID, and interaction parameters to the backend; 3. Video recording – scheduled screen capture stored in OSS for post‑event verification; 4. Orchestration assistance tools – browser plug‑in for one‑click element copying, batch component modification, and DIFF comparison of configuration versions; 5. Jump‑stop prevention – refusing to update when new data is unavailable; 6. Precise cross‑midnight stop – millisecond‑level timing based on system‑time deviation and TP99 metrics; 7. Proxy data source – merging identical user requests into a single backend query and serving cached results to reduce QPS pressure.

Heartbeat reporting flow

Real‑time monitoring linked with pre‑plan actions

The platform also added capabilities such as video‑recording tasks for cross‑midnight validation, orchestration tools to accelerate configuration changes, and a proxy data source to alleviate backend load.

In the final reflection, the team summarized the promotion success, distilled reusable capabilities (heartbeat monitoring, proxy data source, recording service, DIFF tool), and identified improvement areas like reducing communication overhead and standardizing processes.