Why Data Warehousing Is the Boring, Brutal, Beautiful Future of Data
Martyn Rhisiart Jones
A Coruña, 5/12/2025
The modern enterprise data stack looks less like technology and more like cyberpunk plumbing. Pipes are rerouted through pipes. Valves are attached to valves. Layers are placed on top of each other until the whole thing resembles a drawing. It’s like a creation made by a caffeinated child let loose on a whiteboard. Data lakes are here. A data mesh is over there. A lakehouse is balanced precariously on top. It resembles a loft extension built by someone who learnt carpentry from TikTok.
Somewhere, a CTO is still insisting the data lake will “start delivering business value any quarter now.” This is like a man leaving a voicemail for his own kidnapper. He tries to sound calm while absolutely nothing is under control. Big data conferences have become like revivalist churches. They are all shouting and sweating. Promises of salvation are made, just a few more nodes away.
Meanwhile, in the corner, the data warehouse stands silently. Boring. Predictable. Embarrassingly competent. It is like an accountant who can actually tell you where your money went. But this can only happen if you stop screaming about semi-structured log files long enough to listen.
This is the story of how we mistook chaos for innovation. We are finally crawling back to the architecture that always worked.
The Big Lie of Big Data
Big data sold us a fantasy. It suggested that if we just collected enough stuff, intelligence would emerge. This included every click, scroll, sneeze, and packet. Intelligence would emerge by sheer geological pressure. It was the tech equivalent of storing all your receipts in a skip and calling it “forensic accounting.”
Hadoop clusters spread through enterprises like a rumour in a 1980s school playground, fast, incoherent, and ultimately disappointing. Companies didn’t become data-driven; they became infrastructure-driven, endlessly feeding the distributed beast.
My pretend mate Mark Steel might put it like this: “Businesses in this industry set fire to millions in hardware.” They proudly announce the problem is they didn’t burn enough.”
After a decade of worshipping scale-at-any-cost, the only things that actually scaled were cloud bills and existential dread.
Data Lakes: The Digital Junk Drawer
Data lakes promised freedom. A wide-open place to store everything. Neutral. Flexible. Switzerland, but wet.
Instead, they became the garage. Every department dumps their junk there. They swear they’ll sort it “later.” Then, they pretend it isn’t there.
Schema-on-read? More like schema-on-regret.
The ever-reliable Stewart Lee would explain that you simply dump everything into the lake. Yes, everything. Of course, everything. Then, after a long pause, he would explain again that you dump everything into the lake. This time, slower, sadder, and more disappointed in humanity.
A data lake is not a lake. It is a bog of eternal stench, the one from Labyrinth, but with more corrupted Parquet files. It’s the sort of place where scouser Alexei Sayle might cycle through on a rusted bike, yelling about Marxism. At the same time, you desperately try to remember which bucket held the “customer” table.
Lakehouses: A Shed on Stilts in a Swamp
The lakehouse is basically an attempt to rescue a metaphor that was never fit for purpose. It slaps a warehouse interface on top of a swamp and insists everything is now enterprise-ready.
A lakehouse is what you get when someone says, “But what if we put a house on top of the swamp? That’s progress, right?” Sure, Carl. Whatever helps you sleep at night.
It’s a duct-taped compromise that somehow manages to deliver the complexity of a lake with the fragility of a warehouse. It’s technology driven by property development programmes. Nobody is willing to admit the foundations are made of wet cardboard.
Stewart Lee would stare blankly at you for 30 seconds and say, “Yes. Yes, of course. This time, building the house on the swamp will work. This time the water won’t rise.”
Data Mesh: A Workplace Sitcom Gone Wrong
Data mesh is a lovely idea. This is true only if you live in a utopia. It may also work in a parallel universe. In such places, middle managers can define a proper data contract without having a mild breakdown.
In practice, it’s a sociology experiment conducted on production systems.
“Every team should own their data products!” is the tech equivalent of saying “Everyone on the bus could land the plane if required.” It’s optimistic to the point of being dangerous.
Most teams can barely get everyone to update their password without sending a GIF of a crying panda. And now they’re expected to run data pipelines, define schemas, manage governance, and carry the torch of organisational enlightenment?
As Alexei once said about bureaucracy: “It’s not that it goes wrong, it’s that going wrong is the system.” A data mesh is a beautiful theory that collapses instantly under the weight of reality.
The Warehouse: The Adult in the Room
While everyone else was playing buzzword bingo, the data warehouse quietly continued doing the thing that actually works: enforcing structure.
It demands clarity. It forces decisions. It eliminates ambiguity through the radical technique of not allowing nonsense.
Where the lake is all freedom and entropy, the warehouse is discipline and meaning. It is the adult in a room full of people proposing we “just use S3 as a database.”
The warehouse doesn’t chase vibes. It delivers answers, quickly, reliably, consistently. It’s boring in the same way bridges are boring: strong engineering is supposed to be boring.
In a world of tech solutions that behave like experimental theatre, the warehouse behaves like it has read the manual.
Why Warehousing Wins the Next Decade
Here’s the twist modern data architects are slowly rediscovering:
Clean data beats big data.
Structured data beats flexible-but-useless data.
Correct data beats “we’ll fix it later” data.
AI doesn’t want your swamp. It wants sanitised, governed, high-quality information, the kind warehouses are built to deliver.
AI models trained on lake data behave like a drunk uncle at Christmas: loud, confident, and demonstrably wrong.
Warehousing removes the mystery and the misery. It’s the Cristiano Ronaldo of data: it sparks joy by ruthlessly deleting anything that isn’t useful.
The Future: Boring Wins, Again
After a decade of hype cycles and rebrands, we’ve come full circle. Architecture diagrams look like subway maps from dystopian cities. Marketing teams name every new storage combination a “paradigm shift.”
Structure scales.
Chaos doesn’t.
The warehouse, embarrassingly, was right all along.
The next era belongs to companies that choose clarity over chaos. This is bad news for organisations whose data strategy now resembles a Jackson Pollock painting.
The big revelation of the 2020s is that the technology that actually works is… the one we already had.
Try selling that at a tech conference.
Many thanks for reading!
Discover more from GOOD STRATEGY
Subscribe to get the latest posts sent to your email.
GOOD STRATEGY 