How Huawei Tackles Data Silos: Lessons from “Huawei Data Way”

Why Huawei Needs Digital Transformation

Huawei historically built many independent IT systems, each with its own database. This resulted in severe data‑island problems: inconsistent data definitions, duplicated entry across systems, and mismatched records. In the context of digital transformation, data becomes a core production factor, making data governance essential.

Value of an Information Architecture

A well‑designed information architecture does more than support IT implementation; it provides a systematic way to manage corporate data assets and improves overall business efficiency.

Core Elements of Data Governance

Data Asset Catalog

Data Standards Company‑wide definitions of data attributes and rules that must be uniformly followed, establishing a shared understanding of each data element.

Data Model A conceptual abstraction of real‑world features, capturing key business entities and their relationships.

Data Distribution Defines the source of data and its flow across processes and IT systems. The primary data source is certified by a data‑management organization and serves as the enterprise‑wide authoritative source.

Building an Architecture Based on Business Objects

Designing Around Business Objects

Business objects represent the most important people, events, and things in a domain. They carry critical information and act as the key linkage between business and IT, forming the foundation for information, business, application, and technology architectures.

Implementation Steps

1) Conceptual Data Model

Provides a high‑level overview of key enterprise data entities and their relationships. The main activities are:

Identify core business entities (e.g., Customer, Product, Order).

Define high‑level relationships (e.g., Customer places Order, Order contains Product).

Document the model in a diagram that can be shared with business stakeholders.

2) Logical Data Model

Refines the conceptual model by specifying each entity’s attributes, data types, relationships, and constraints. This model focuses on business logic and is independent of any specific DBMS.

Typical tasks:

Define attribute names, data types (e.g., VARCHAR, DATE), and nullability.

Specify primary keys, foreign keys, and unique constraints.

Model many‑to‑many relationships using associative entities.

Tools such as PlantUML or Drawio can be used to create the diagram.

3) Physical Data Model

Translates the logical model into concrete database structures tailored to a specific DBMS. This includes:

SQL DDL statements to create tables, indexes, and constraints.

Storage parameters (e.g., tablespace, partitioning).

Performance considerations such as indexing strategy and column compression.

The physical model is the implementation blueprint for developers and DBAs.