TMI / COMPARISONS
DATA WAREHOUSE
VS DATA FOUNDATION.
Short answer. A data warehouse is a place to store and model structured records from systems that already produce them. A data foundation is the whole job of making what a company knows usable — which in most established companies includes paper, disconnected systems and knowledge that was never in a database at all. A warehouse can be part of a foundation. It is rarely the whole of one.
WHAT EACH ONE
ACTUALLY IS.
DATA WAREHOUSE
A data warehouse centralises structured data from source systems so it can be queried and reported against without hammering the systems themselves. It is well-understood technology, it scales, and for a company whose data already lives in databases it is often exactly right.
It assumes its inputs. A warehouse fed by two systems that disagree about a customer stores the disagreement efficiently.
DATA FOUNDATION
A data foundation is the work of getting a company to the point where its records can carry a decision: aggregating them, connecting the systems, cleaning what is dirty, digitising what is on paper, capturing what only exists in a long-serving employee's head, and agreeing a source of truth.
It is defined by the outcome rather than by the technology, so what it contains varies enormously between companies. In an industrial business a large part of it is frequently not technical at all.
DATA WAREHOUSE
WHEN…
/DATA FOUNDATION
WHEN…
CHOOSE DATA WAREHOUSE
- The data already lives in systems and is already reasonably clean
- Reporting is slowing the production systems down
- There is genuine volume, or several databases that need to be queried together
- Someone in-house can own the modelling
CHOOSE DATA FOUNDATION
- Records that matter are on paper, in email or in people
- Two systems describing the same customer do not agree
- The most valuable knowledge in the company has never been written down
- Nobody is sure which source wins when sources conflict
Neither column is a sales argument. Both appear on this page because both are correct somewhere, and a comparison in which one side never wins is an advertisement.
HOW TO DECIDE
THREE QUESTIONS
THAT SETTLE IT.
- 01
What proportion of what you need is already in a database?
The honest answer to this decides the question. Below about half, a warehouse is premature infrastructure. - 02
If the warehouse existed tomorrow, could you trust what it contained?
If the answer is no, the work is upstream and buying storage postpones it. - 03
What is the first decision this is meant to improve?
Naming one keeps a data programme from becoming an infrastructure project with no finish line.
TMI builds data foundations and will build or specify a warehouse where that is the right component. The order matters more than the choice: storage before agreement produces an organised version of the same confusion.
STILL NOT SURE WHICH