Low-Code Process Engine Architecture and Implementation Using LiteFlow

Background: As membership business grows, a middle‑platform system for products, orders, and rights is formed, supporting rapid business innovation. However, configuration backends need to call common interfaces, leading to repetitive custom development.

To address this, we shifted from traditional waterfall development to a workflow engine approach. This article shares our practice.

Solution Design: General interfaces are designed as components assembled via a pipeline. Business configuration functions require no code, following a workbench model (see diagram). The combination of "workbench, component, rule" achieves high decoupling and flexible orchestration.

Workbench provides a unified context, a repository for inputs and intermediate results, and two‑way data exchange between components.

Components are independent workers with clear responsibilities, high cohesion, low coupling, and flexible scheduling through the workbench.

Rules define execution order via simple configurations, offering readability and flexibility. They support serial, parallel, conditional, loop, and exception handling arrangements.

Technical Architecture: Among many workflow engines, we chose the open‑source Chinese framework LiteFlow for its lightweight, simple configuration, rich features, and low trial cost.

LiteFlow Architecture Diagram:

LiteFlow Features: Rule expressions (EL) use concise DSL syntax to define component execution flow. Examples include serial arrangement A -> B -> C , parallel execution, conditional branching, loops, and exception handling.

Advantages of rule expressions: decompose complex processes, flexible composition and reuse, dynamic order adjustment, and visual configuration.

System Implementation: Generic components provide core capabilities such as network request, database operations, and activity rules, which can be assembled to meet business scenarios.

Component Example: Network request component consists of request configuration, fixed request framework, and custom parameter parsing.

Scenario Example: Using workflow orchestration to configure and approve marketing activities, reducing development cost and enhancing system extensibility.

Conclusion: The low‑code workflow engine model suits complex business logic, abstracting code decisions and branching, improving modular cohesion, reusability, and reducing development and maintenance costs. It applies not only to complex C‑end processes but also to backend configuration efficiency, with future exploration of more scenarios.

Future Plans: Build a backend workflow engine platform with capabilities such as visual process management, high‑performance execution engine, state tracking and replay, and intelligent decision support.