Didi Internationalization Site Building: Application Modeling and Cloud‑Native Delivery Framework

Background: Didi's international travel platform shares experiences on improving data‑center construction efficiency through three aspects—efficiency in planning, delivery, and refactoring—aiming to build an application‑centric automated delivery framework.

Goal analysis: The overall goal is to establish a standardized management model describing all workloads of an application, enabling standardized and high‑efficiency delivery in hybrid‑cloud environments.

Application model: Emphasizes separation of concerns among infrastructure, development, and operations teams; requires a platform‑agnostic, highly extensible model that can describe compute, storage, middleware, and operational features, supporting multi‑cloud deployment.

Delivery framework: Automates delivery processes for diverse dependencies, ensuring correct ordering and providing a highly extensible integration method that abstracts underlying platforms.

Solution ideas & challenges: Proposes a standardized application description based on the OAM model, managing the full lifecycle of dependencies, handling resource lifecycle, and addressing challenges such as template flexibility, resource association, and configuration changes.

Technical implementation: Uses OAM to define components (business‑focused) and traits (operations‑focused), builds an OAM‑based application description, and adopts KubeVela (an OAM‑compliant cloud‑native delivery platform) with CUE language for extensible modules.

Component abstraction & UI standardization: Introduces a component market where services (e.g., MySQL, Redis) publish definitions; UI‑Schema is generated dynamically to render configuration forms, supporting fixed, dynamic, and automatically generated values.

Delivery engine & workflow: Integrates the delivery engine into existing CI/CD pipelines, defines stages and tasks, and provides a customizable workflow that can orchestrate arbitrary dependency ordering, conditional logic, and status monitoring.

Observability & status management: Chooses a polling‑based approach to monitor component delivery status, leveraging KubeVela’s declarative reconciliation to keep the workflow up‑to‑date.

Summary & outlook: The application center has significantly improved delivery efficiency for Didi’s international travel services, and future work will focus on finer‑grained workload management and multi‑active data synchronization.