Every builder gets a defining project eventually — the one that tests whether the methods are real or just things you say. Mine was a monolithic Drupal 7 application, years past its build date, millions of nodes deep, that needed to become Drupal 10 while the site kept publishing. Sole technical lead. The client stays unnamed by agreement; the case study and the war-story version cover the engineering. This is the part those leave out: what it was like to carry.

A project that size is a season, not a sprint — and the middle of it is the longest portage I’ve walked. There’s a stretch in any long carry where the put-in is too far behind to see and the next lake is a rumor, and morale lives or dies on one thing: knowing which trip you’re on. The counts saved me there, and not only technically. Reconciling every entity type every run meant that on any given morning, however endless it felt, I could say exactly how much gear had crossed and how much remained. The number was the antidote to the fog. I’ve never once regretted over-measuring a project; I’ve watched plenty of people drown in under-measured ones.

The other thing the big one teaches is respect for accumulated intent. Millions of nodes isn’t just data — it’s a decade of some organization’s decisions, workarounds, and institutional memory, encoded in reference fields by people who never wrote down why. A migrator’s real job is archaeology with a deadline: preserve intent you can only infer, and stay humble about the difference between “this looks vestigial” and “this is load-bearing.” Cut confidently and you find out at the worst moment, in front of the long table.

It ended the way the method promised: cutover measured in minutes, counts matched, nothing lost. The pride isn’t the size. It’s that a promise that specific — every record, intact — turned out to be keepable.

Next: Guardrails — what the big one left behind.