How a Data Platform Actually Grows
From a BI tool pointed at a replica to a platform other teams build on. Each rung appears only when a problem forces it — read the problem first and decide what you would change before you look at what the rung turned out to be.
The reference architecture diagram everybody has seen is the top of this ladder with the bottom removed. Adopted all at once, it is not architecture — it is a dozen components you now operate, most of them solving problems you have not got, each one adding a way for the data to be wrong.
Every rung below costs something, and the cost is only worth paying against the problem stated above it. If you cannot name the problem in your own platform, the rung is not yours yet.
- Rung 1 · One product, one database, a few analystsThe problem that forces thisAnalysts want to answer questions and there is nowhere to run the query except production.What changes
Point a BI tool at a read replica and save the queries.
What it buysAnswers today, with no pipeline, no second copy and no operational commitment.
What it costsAnalytical load still competes for the replica, history is whatever the operational schema happens to keep, and nothing joins to a second source.