GLOSSARY

Data Warehouse

A data warehouse is a centralized analytical database — Snowflake, BigQuery, Redshift — holding integrated, modeled data for BI and reporting.

Last updated:

Quick answer
A data warehouse is a central repository of structured, cleaned, and modeled data optimized for analytical queries. Modern cloud warehouses — Snowflake, BigQuery, Redshift, Databricks SQL — separate storage and compute, scale elastically, and are the default destination for most enterprise analytics. Kimball-style dimensional modeling, paired with dbt, is the predominant practice in the modern stack.

WHAT IT IS

Leading cloud warehouses include Snowflake, BigQuery, Redshift, Databricks SQL, and Microsoft Fabric. Data is modeled dimensionally (Kimball star schema) or normalized (Inmon) or in a Data Vault, then exposed to BI tools through a semantic layer that enforces metric definitions.

HOW IT WORKS

Where an operational database (OLTP) is tuned for many small transactions, a warehouse (OLAP) is tuned for few large queries — columnar storage, partition pruning, and result caching. Modern warehouses blur the line with lake access (external tables), streaming ingestion, and ML runtimes in the same engine.

WHEN TO USE

Choose a warehouse-centric architecture when reporting workloads dominate, when data sources are mostly structured, or when governed BI is the primary consumer. Choose a lakehouse when ML and semi-structured workloads are also first-class.

RELATED

SOURCES

Related questions.

What is a data warehouse?
A data warehouse is a central repository of structured, cleaned, and modeled data optimized for analytical queries. Modern cloud warehouses — Snowflake, BigQuery, Redshift, Databricks SQL — separate storage and compute, scale elastically, and are the default destination for most enterprise analytics.
How is a data warehouse different from a database?
A transactional database is optimized for many small writes and reads from an operational application (OLTP). A warehouse is optimized for large analytical queries across historical data (OLAP). The access patterns are fundamentally different, which is why both exist.
What data modeling is used in a warehouse?
Kimball's dimensional modeling (star and snowflake schemas, fact tables and dimension tables) remains the default for BI-first warehouses. Inmon's normalized 3NF approach suits enterprise integration layers. Data Vault 2.0 is used in large, change-resistant environments. dbt has made Kimball-style modeling the predominant practice in the modern stack.
When should a company build a warehouse versus buy one?
Almost always buy. Snowflake, BigQuery, Redshift, and Databricks SQL offer cloud-native warehouses that are faster to stand up and cheaper to operate than self-managed alternatives at any reasonable scale. Building your own warehouse is a decision that requires an extraordinary reason.
How does NUUN Digital build warehouses?
We standardize on cloud warehouses (Snowflake, BigQuery, Redshift, Databricks SQL), model with dbt, and publish a semantic layer so every metric has one owner and one definition. Warehouse builds ship in ninety-day increments, with value delivered each quarter.

Need this term in action?

Book a call