Data Warehousing Overview

Principles

Definition

Data warehousing is the systematic process and methodological framework for collecting, integrating, and storing data from diverse, often heterogeneous sources into a unified, large-scale repository specifically structured and optimized to support analytical querying, reporting, and business intelligence activities. As an academic concept and research field, it investigates the architectures, techniques, technologies, and management strategies necessary for effective design, implementation, and utilization of such repositories to enable data-driven decision-making and strategic analysis by providing a comprehensive, historical view of organizational data.

Ontological type

Architectural Models

ETL and Data Integration

Design Methodologies

Key developments

Key Figures

Dimensional Modeling & Views era

Inderpal Singh Mumick [1] was a seminal figure in data warehousing during the Dimensional Modeling & Views era (1999-2005), with affiliations spanning universities and research centers active in that period. His key contributions include the paper Selection of views to materialize in a data warehouse [2], published in 2004, which advanced how analytic intents were preserved while balancing ETL overhead and query latency. This work catalyzed the standardization of selective materialization practices and informed subsequent tooling and optimization efforts across the era [2]. Thus, Inderpal Singh Mumick's contributions helped align business semantics with dimensional models and accelerated adoption of view-based optimization in warehouse architectures during 1999-2005 [1].

Distributed Real-Time Warehousing era

Ning Zhang [1] emerged as a leading figure in distributed real-time warehousing during 2006-2014, with academic ties at University of California, Berkeley [2] and Cornell University [3]. Her notable contribution in this era is the Hive paper, Hive - a petabyte scale data warehouse using Hadoop [4], published in 2010, which demonstrated how Hadoop-based storage and parallel processing could scale analytics to massive datasets and support batch workloads. Ning Zhang [1] helped advance low-latency analytics on commodity infrastructure by promoting partitioned storage and parallel query workflows that align with stream-like ingestion patterns. In this era, Ning Zhang [1] demonstrated how partitioned storage and parallel query pipelines enabled near-real-time analytics on commodity infrastructure.

Cloud-Native Lakehouse era

Ali Ghodsi and Matei Zaharia of Databricks anchor the cloud-native lakehouse era, with Ghodsi promoting the lakehouse abstraction and Delta Lake as an ACID, schema-enforcing layer atop object stores and Zaharia underpinning it with scalable Spark-based analytics. Bill Inmon and Ralph Kimball remain representative pioneers of data warehousing, whose warehousing principles—enterprise integration and dimensional modeling—inform governance, metadata catalogs, and policy-driven design in lakehouse architectures. Werner Vogels of AWS has advocated decoupled storage and elastic compute in the cloud, shaping practical deployments of lake storage, managed ETL, and on-demand compute that underlie cloud-native platforms. Michael Stonebraker and other HTAP researchers also influence the era by stressing converged analytics and unified data management that motivate converged query engines and hybrid deployment models.