What’s Next for Data Warehouses? From History to Future Trends

In this live broadcast titled “From the History of Data Warehouses to Future Technology Trends”, Huawei Cloud’s EI DTSE evangelist and chief architect of GaussDB(DWS) Zeng Kai discussed the origins, evolution, and future directions of data warehouses.

Where did data warehouses come from? In the 1970s, E.F. Codd introduced the relational model, and MIT proposed the theory of separate analytical systems. In the early 1980s, W.H. Inmon began research on “record systems”, “original data”, and “decision support databases”. In the early 1990s, Inmon published the classic “Building the Data Warehouse”.

Since the 1990s, data warehouses have flourished, producing many products. Using Inmon’s definition, a data warehouse is a subject‑oriented, integrated, relatively stable collection of historical data that supports managerial decision‑making. Its four main characteristics are:

Subject‑oriented – data is classified by business subject for easier understanding.

Integrated – data is processed and unified with a common structure and coding before use.

Non‑volatile – historical data is stored with little or no updates.

Time‑variant – data accumulates over time, preserving historical trends.

Traditional technical architectures can be grouped into three models: Shared Everything, Shared Storage, and Shared Nothing. The Shared‑Nothing architecture, with its excellent scalability, has become the mainstream for high‑performance data warehouses.

Logical layer architecture typically consists of four tiers: Operational Data Store (ODS), Data Warehouse Detail (DWD), Data Warehouse Service (DWS), and Application Data Service (ADS). ODS stores raw source data; DWD refines data for quality and dimensional modeling; DWS aggregates data for analytical domains; ADS provides data products and analytics for downstream consumption.

Future trends point to massive, diverse, and real‑time data, with business decisions requiring instantaneous insights. Data volume grows ~30% annually, doubling every three years, and data types expand beyond logs to images, remote‑sensing, and files.

Huawei Cloud GaussDB(DWS) evolution spans 12 years: research began in 2011, first release in 2014, large‑scale commercial use since 2017, and over 1,700 global customers today. In 2022, real‑time and IoT data warehouses were launched. In March 2023, a cloud‑native data warehouse was announced, featuring three‑layer compute‑storage separation, extreme elasticity, lake‑warehouse integration, AI‑data fusion, and performance comparable to traditional warehouses.