Design and Implementation of the JinDao Online Settlement Platform

The JinDao settlement platform was created to address the rapid growth of JD Technology’s business and the resulting demand for online fund settlement, overcoming the high labor cost, long processing time, coordination difficulty, and low accuracy of traditional offline settlement methods.

Key pain points include complex calculation rules, slow response to new settlement models, high risk of manual errors, lack of multidimensional operational analysis, and high resource consumption.

The platform positions itself as a bridge for finance‑business integration, supporting cross‑platform settlement through a modular architecture based on Domain‑Driven Design (DDD) that separates the system into four main domains: clearing, settlement, accounting, and reporting.

Core terminology such as clearing , settlement , billing , allocation , cumulative amount , and reset are defined to standardize the processing flow.

System architecture utilizes a sharding‑by‑table strategy with data synchronized to Elasticsearch for reporting, while ensuring monetary accuracy by avoiding reliance on ES for aggregation. A custom shardingTask component abstracts the split‑process‑merge workflow, handling massive data slices via MQ and multi‑node parallel processing.

Sequential settlement is achieved by tagging records with a unique, ordered serial number, processing them through a staged pre‑settlement workflow, and merging results while supporting both ordered and concurrent execution for high‑traffic scenarios.

The settlement process consists of four steps: (1) defining settlement schemes, (2) configuring settlement models (including billing, clearing, settlement, and downstream rules), (3) executing settlement tasks that generate settlement orders and trigger confirmation workflows, and (4) finalizing payments through financial channels.

Configuration covers basic information, settlement cycles, document handling, clearing, settlement, and downstream integration, enabling a one‑stop setup for operations teams.

Billing and allocation models support configurable rules such as proportional, sequential, and fixed‑amount distribution, with two‑level allocation and budget‑based controls, meeting diverse business scenarios.

Real‑world case studies include micro‑electric business (commission, performance assessment, WeChat promotion fees), white‑bar business (marketing fee collection), and payment business (external institution coupon fees), demonstrating the platform’s ability to process over 50 million orders daily and settle more than 13 million RMB in value.

Four settlement modes—commission, performance assessment, technical service, and product marketing—cover current business needs, with the architecture designed to accommodate future extensions.