How JD’s Low‑Code Rule Engine Cut Delivery Time to Half a Day for Billion‑Order Peaks
This article explains how JD Supply Chain’s self‑built low‑code rule‑engine platform separates business logic from code, supports ultra‑high concurrency, enables non‑technical users to configure rules, and reduces a typical 10‑day delivery cycle to as fast as 0.5 day across dozens of core projects.
1. Introduction
Rule engines are common technical tools in the industry, with open‑source options like Drools that separate volatile business rules from application code, improving delivery efficiency and reducing maintenance costs. However, traditional engines struggle with massive traffic, non‑technical user participation, and lack standardized, scalable architectures.
JD Supply Chain R&D built a low‑code rule‑engine platform called HaiNa, targeting both business and development roles. The platform’s core rule engine enables high‑concurrency internet services and provides a standardized, extensible framework.
Within four months, the platform opened over 100 extension points, extracted more than 400 business rules, supported 14+ major JD projects, and allowed product managers and ISVs to self‑service requirements with an average delivery time of 0.5 day.
Long‑term, the platform expands development functions to more roles safely, laying the foundation for scalable ecosystem delivery.
2. JD Fulfillment Application
The HaiNa platform is already used at scale in fulfillment, with about 20% of requirements handled directly by business roles, projected to reach 40%.
A narrative illustrates a typical workflow: a junior engineer receives a new requirement to package high‑end cosmetics in brand‑specific gift boxes, evaluates the implementation, and faces tight deadlines. Meanwhile, senior architects identify commonalities across many requests, recognizing that 40% of demands can be expressed as “when certain business conditions are met, execute specific actions,” a perfect fit for rule‑engine automation.
3. Core Principles
The platform offers a visual designer where business users compose rules as flowcharts; the backend automatically generates rule‑engine description files, which are synchronized to applications via SDKs. Execution occurs at registered extension points, with business models providing data context and technical rules handling low‑level operations.
Key features include:
Lightweight integration : only an SDK and annotations are needed, enabling one‑day integration for legacy systems.
High performance : supports over a billion orders per day, with rule execution in nanoseconds, proven during JD’s 11.11 promotion.
Separation of development and business : business users configure rules without touching code, while developers maintain the underlying engine.
Visual composition : rules are built like flowcharts, offering a WYSIWYG experience.
Technical asset accumulation : registered business rules become reusable assets, accelerating future deliveries.
Images illustrate the platform architecture and rule‑flow diagrams:
4. Summary
The HaiNa low‑code rule engine empowers business roles with visual rule composition, while developers integrate via a lightweight SDK. This decoupling shortens delivery from ten days to half a day, supports billion‑order peaks, and creates a reusable technical asset library for continuous improvement.
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
21CTO
21CTO (21CTO.com) offers developers community, training, and services, making it your go‑to learning and service platform.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
