Executive steering
Reliable historized dashboards.
What is a data warehouse? Analytical storage explained
A data warehouse is a database optimized for analysis: it centralizes data from multiple operational systems, historized and modeled for reporting and BI. Unlike OLTP that serves today's transactions, the warehouse answers “what happened?” over large volumes, often via analytical SQL or BI tools.
In one sentence
A data warehouse centralizes business history so you can analyze without slowing production.
Key points
- Separated from operational DBs so they stay fast.
- Star/snowflake or wide-table models depending on approach.
- Fed by ETL/ELT from CRM, ERP, web, etc.
- Cloud warehouses (BigQuery, Snowflake, Redshift) lowered the barrier.
Term at a glance
- Data warehouse
- Enterprise data warehouse · EDW · Cloud data warehouse
- English term
- Data warehouse
- Domain
- Data
- Category
- Analytics
- Level
- Intermediate
What does “data warehouse” mean exactly?
Classic idea (Inmon/Kimball): a consolidated analytical source of truth. Cloud warehouses now scale storage and compute separately.
You load facts (sales, events) and dimensions (customer, product, time).
For an SME, a light warehouse can start as analytical Postgres or a managed warehouse once Excel exports explode.
How do you stand up a data warehouse?
- 01
Map sources
CRM, ERP, e-com, logs—quality and owners.
- 02
Choose the platform
Cloud warehouse or Postgres by volume/budget.
- 03
Model for business questions
KPIs first, not every column “just in case”.
- 04
Automate ETL/ELT
Pipelines, quality tests, documentation.
Concrete data warehouse example
A clinic chain in Quebec consolidates appointments, billing, and marketing campaigns in a warehouse. Executives track no-shows and CAC per clinic without Monday Excel extractions.
What is a data warehouse for?
Self-service BI
Business explores without blocking OLTP.
Data science
Clean sets for predictive models.
Compliance reporting
Auditable aggregates and traces.
Pros and cons of data warehouses
- Analyze without hitting prod
- Coherent history
- SQL familiar to analysts
- Cloud scalability
- Cost if poorly governed
- Upstream quality required
- Latency vs strict realtime
- Modeling skills needed
Data warehouse vs operational DB (OLTP)?
| Data warehouse | Operational database (OLTP) | |
|---|---|---|
| Goal | Analyze history | Run transactions |
| Workload | Heavy reads, aggregates | Frequent writes, low latency |
| Model | Denormalized / dimensional | Normalized for integrity |
| Users | Analysts, executives | Business applications |
Why a data warehouse matters for a Quebec SME
When every department has “its” Excel truth, decisions diverge. An aligned warehouse cuts reconciliation meetings and speeds steering—especially across sites.
FAQ
Warehouse = data lake?
No. Lakes store raw polyglot data; warehouses serve structured analytics. They often coexist.
Snowflake on day one?
Not always. Validate business questions; well-designed Postgres can suffice.
Realtime?
Warehouses are often near-real-time; strict streaming may need other pieces.
Who owns the data?
Define business data owners—tech alone is not enough.
Related terms
Sources and references
Still living in diverging spreadsheets? We can scope a pragmatic warehouse.
Talk about data warehousing