How a Data Middle Platform Transforms Business: Design, Architecture, and Modeling Insights

1. About the Data Middle Platform

Many public articles now discuss the concept of a data middle platform . Companies like Supercell and Alibaba have achieved remarkable results by adopting a "small front‑end, large middle‑end" architecture. Our company, Zhenkunxing, is also planning its own data middle platform.

The data middle platform integrates data warehouse, data lake, and big‑data platforms into a comprehensive ecosystem. While traditional data warehouses mainly serve reporting needs for a limited group of analysts, a data middle platform serves a broader audience, initially business middle‑ends and later external partners, requiring a full‑stack data service ecosystem.

Its architecture includes storage and compute platforms, as well as development IDE, workflow, data sync, external API services, data asset management, OneID, and agile reporting tools.

2. Data Warehouse Architecture Design

Based on the above thinking, we propose an initial model of the data middle platform, illustrated below:

The layered design consists of:

ODS layer : storage layer mirroring business databases, facilitating traceability and supporting data desensitization during synchronization.

DWD layer : detailed data layer where data is standardized (e.g., code conversion, metric formatting).

DWT and DIM layers : abstracted subject‑area tables and common dimension tables.

DWA layer : aggregated data covering common dimensions and metrics.

Both detailed and aggregated data are needed for various application scenarios, so the data flow follows the diagram shown below:

3. Data Model Design

We adopt Ralph Kimball’s dimensional modeling method. After outlining the business middle‑end blueprint, we identify domain areas and their sub‑domains, grouping related systems accordingly. The domain diagram is shown below:

Due to confidentiality, we illustrate only the supplier‑management domain as an example:

4. Summary of Data Middle Platform Construction

The primary value of a data middle platform is to empower the business middle‑end. It focuses on two aspects:

Business Data‑ization : enabling data reuse across business units, real‑time tracking of business events, and stronger data processing capabilities.

Data Business‑ization : turning data into services such as reporting, custom analysis, self‑service analytics, data‑driven operations, and AI‑enabled external services.

A mature data middle platform amplifies data value, drives new business innovation, supports company growth, and can eventually offer data services to external partners.

In summary, a well‑designed data middle platform can help a company achieve a strategic advantage by better supporting business development.