How Modern Data Warehouses Evolve: Insights from Huawei’s GaussDB(DWS) Chief Architect

Data warehouse development has long been a hot topic, and recent technological advances continue to drive its evolution.

We invited Zeng Kai, chief architect of Huawei Cloud Data Warehouse and member of the China Computer Federation’s Database Committee, to share his perspective on data warehouses.

Zeng earned his bachelor's degree at Zhejiang University and his Ph.D. at UCLA, conducted post‑doctoral research at UC Berkeley’s AMPLab, and has published multiple CCF‑A database papers, winning SIGMOD best‑paper and best‑demo awards.

He explained that databases were created alongside computers, but data warehouses only appeared after the 1980s when growing data volumes and analytical demands required dedicated analytical workloads. He described the evolution from descriptive warehouses to exploratory, operational, and now intelligent warehouses that extract business value from data.

According to Zeng, the future direction of data‑warehouse products can be summarized by “fusion”: the integration of traditional warehouse technology with cloud computing, batch and streaming, lake and warehouse, data and AI, and transactional and analytical processing (HTAP).

Traditional warehouse technology + cloud computing : cloud‑native, serverless architectures with compute‑storage separation provide extreme elasticity, lower cost, and higher resource utilization.

Batch + streaming : real‑time analytics for scenarios such as fraud detection, marketing, and credit approval require immediate analysis of newly generated data.

Lake + warehouse : a unified lake‑warehouse combines the high performance of warehouses with the low cost of data lakes, supporting hot, warm, and cold data tiers and handling structured, semi‑structured, and unstructured data.

Data + AI fusion : integrating data‑platform capabilities with AI platforms enables complementary strengths, allowing data‑warehouse management to feed high‑quality data into ML pipelines.

TP + AP (HTAP) : using HTAP builds systems that support both transactional and analytical workloads, reducing cost and simplifying operations.

When asked why he chose GaussDB(DWS), Zeng highlighted his long‑standing focus on databases and distributed systems, the product’s ten‑year evolution from Huawei’s 2012 lab, its service to over 1,700 customers, and its strong academic presence through publications and conference participation.

Looking ahead, GaussDB(DWS) will continue to invest in cloud‑native serverless architecture, real‑time analytics, HTAP mixed workloads, and deeper integration with the surrounding ecosystem, including lake‑warehouse unification, data‑intelligence fusion, and AI‑enabled analytics.

GaussDB(DWS) 3.0, slated for release at the end of March, retains the classic performance and reliability while adding minute‑level elastic logical clusters, fine‑grained resource isolation, linear scalability, enhanced lake‑warehouse queries, and seamless AI pipeline integration.