Data Governance Practices in Meituan Delivery: Architecture, Standards, and Security

Data assets have become a core competitive advantage for enterprises, but without proper data governance, investments can be wasted. This article, authored by Meituan’s delivery data governance team, systematically explains their data governance concepts, objectives, timing, and implementation methods.

Background : In the era of big data, many companies treat data as a strategic asset. However, neglecting governance leads to inconsistent data, poor quality, security risks, and ineffective model usage.

1. Understanding Data Governance : It is a framework for managing, evaluating, guiding, and supervising large‑scale data, ensuring risk control, compliance, performance improvement, and value creation. It covers both existing (stock) data and incoming (incremental) data.

2. Goals : The aim is not the governance itself but to enable strategic objectives—improving data production, management, and usage, standardizing processes, and ensuring data safety and consistency for seamless sharing.

3. When to Start : Governance should be a continuous effort across all data‑warehouse phases (prototype, iterative, and consolidation). The focus and scope differ per phase, but the initiative never ends.

4. How to Conduct Governance :

4.1 Define Standards & Improve Quality : Establish business, technical, security, and resource‑management standards to guide metric definitions, modeling practices, data‑security controls, and resource budgeting.

4.2 Implement : Apply the standards in three steps—(a) clean up existing data models, (b) enforce metadata governance, and (c) strengthen security governance.

4.1.1 Business Standards : Define metric ownership, standardization processes, and a metric‑management committee to oversee the full lifecycle of metrics.

4.1.2 Technical Standards : Adopt a four‑layer warehouse architecture (operational, foundational fact, intermediate, application) with dimensional modeling, and enforce modeling, metadata, and lifecycle rules for each layer.

4.1.3 Security Standards : Implement data classification, role‑based access, privacy protection, and audit mechanisms to ensure data cannot be leaked or misused.

4.1.4 Resource Management Standards : Abstract tenants, resources, and project groups, assign responsibilities, and map data assets to appropriate resources.

4.2 Metadata Governance : Build organization, processes, and tools to standardize metric definitions, construct comprehensive technical and business metadata, and provide services (e.g., OneService) for table/field queries and lineage.

4.3 Security Governance : Protect sensitive data through encryption, key management, and row‑level access controls, and ensure secure data sharing.

5. Tool Overview :

Wherehows (Data Map) : Enables users to search, understand, and trace data assets, linking business and technical metadata, and supports Q&A comments.

QuickSight (Data Visualization) : Provides self‑service dashboards and data‑set management, allowing users to perform second‑level processing and visualization without developer involvement.

6. Summary & Outlook : After three governance phases, Meituan achieved standardized data, flexible architecture, enhanced security, and a full metadata pipeline. Future work will focus on further security, resource optimization, and leveraging emerging database technologies to improve data usability and reduce user costs.