Water utility · where it starts
Water comes into a city from everywhere — a river swollen with snowmelt, a network of deep wells, seasonal springs that dry up in summer, sewage from households and industry, and plain rainfall. Different sources, none of them clean enough to be used in households, companies and industry.
In data terms
Data arrives exactly like this — scattered across dozens of disconnected systems: sales apps, sensors, spreadsheets, third-party feeds. Each is a silo with its own format and quirks. Everything in this topic exists to do one thing: break down those silos and turn fragmented sources into one unified, trustworthy supply.
Water utility · the raw intake
The first stop is the raw intake basin. Dirty and untreated — everything from every source ends up here before anyone has decided what to do with it. The water is stored, separated by source, and only the big chunks — like whole bicycles — are removed.
In data terms
A data lake is a catch-all landing zone where raw data from every source is dropped in fast, before it's cleaned or organised. Capture everything now, sort it out downstream — so you never lose data you didn't yet know you'd need.
The philosophy is: messy-but-complete.
Water utility · the pipe network
Pipes connect every stage — from intake to the treatment facility for cleaning the water, all the way to the clean water reservoirs. The pipe network makes up the whole water utility system: it's what connects everything.
In data terms
A data pipeline is an automated route that Extracts data from each source and Loads it into the data warehouse. The equivalent of the treatment facility is data Transformation. Those three steps are called ELT. A single company may run hundreds at once — one pulling sales every midnight, another streaming website clicks in real time — all without anyone lifting a finger.
Data world · the data warehouse
The data warehouse is the place where everything happens. It contains not only the data but also the pipelines and the transformations of the data. It is organised, structured storage, built so data can be reliably saved, transformed, combined, found, and retrieved. Every time a dashboard pulls last month's revenue, it reads from one.
What this means
This is where silos actually get broken, and manual stitching-together disappears.
Data world · data marts
From the warehouse, data (or water) is blended for purpose into service stores — homes, hospitals, industry each get the formulation they need. Data marts do the same: purpose-shaped subsets tuned to one team's questions, so the finance analyst and the operations team aren't fighting over the same tables.
What this means
People get data ready for their job, not a firehose.
Data world · the data catalogue
The water operations centre displays a map of every pipe, valve, and tank. Without it, maintenance crews can't find the right valve in an emergency.
What this means
A data catalogue is that map for data — what data exists, where it lives, who owns it, what it means, when it was last refreshed. It isn't the data itself; it's the index to it.
The difference between "we probably have that somewhere" and "here it is, here's the owner, here's how fresh it is." It kills knowledge silos.
Data world · data lineage
When contamination is detected, engineers trace the water back: which source, which treatment stage, which reservoir. Data lineage is the full record of where each piece of data came from and every step it passed through. When a report looks wrong, you follow the line backwards to the exact source and transformation that introduced the error.
What this means
It turns debugging from guesswork into a trace — and it's what lets a regulated company prove data was handled correctly (audit & compliance).