Six Years in a Warehouse Taught Me S3 Storage Classes
Before I taught process improvement, I spent six years as a warehouse keeper. The first time I read the S3 storage classes documentation, I recognized the problem immediately — it's inventory rotation with different words. Here's the mapping, and the one place where the warehouse analogy stops working.
Before I taught process improvement, I spent six years as a warehouse keeper across three companies.
The first time I opened the Amazon S3 storage classes documentation, I had a strange feeling. Not confusion — recognition. I had solved this exact problem before, with a forklift instead of a console.
The warehouse problem
Every warehouse keeper learns the same lesson in the first month: not all inventory deserves the same spot.
The items you pick twenty times a day go at the front, at waist height, closest to the door. The ones you touch twice a month go to the back. The ones you keep only because an auditor might ask for them in three years go to the highest shelf, in a sealed box, and getting them down is a whole operation involving a ladder and somebody's afternoon.
Nobody taught me this as theory. It emerged because floor space near the door is the most expensive thing in a warehouse, and wasting it on slow-moving stock is how you end up walking kilometers a day.
Now look at how S3 organizes storage.
S3 Standard is the shelf by the door. Immediate access, no retrieval charge, no minimum stay. Most expensive per unit of space. You put active data there for the same reason you put fast-moving SKUs there.
S3 Standard-IA and One Zone-IA are the back of the warehouse. Cheaper to store, but you pay a retrieval fee every time you go get something, and there's a 30-day minimum — you can't put a box there and take it out the next morning without paying for the full month anyway.
The Glacier classes are the high shelves. Glacier Instant Retrieval still gets you the box in milliseconds but holds it for a 90-day minimum. Glacier Flexible Retrieval makes you wait minutes to hours. Glacier Deep Archive is the sealed container in the corner with a 180-day minimum and retrieval measured in hours, and it is the cheapest storage AWS offers.
S3 Intelligent-Tiering is the part that genuinely didn't exist in my warehouse. It watches how often each object is actually accessed and relocates it between tiers automatically, charging a small monitoring fee per object. In warehouse terms: an invisible assistant who walks the aisles every night and quietly moves the slow movers to the back.
I would have given a year of my life for that assistant.
Why the mapping is useful
Because the mistake is the same mistake.
In the warehouse, the classic error was putting everything at the front "just in case," and then running out of prime space in six months. In S3, the classic error is leaving everything in Standard because it's the default, and then wondering why the storage bill climbs every quarter.
The opposite error is also identical. In the warehouse, someone decides to be efficient and sends half the inventory to the high shelves — including items that turn out to be needed weekly. Now every pick costs a ladder and twenty minutes. In S3, you move data to Glacier to save money, and then discover you're paying retrieval fees constantly on something you actually read often, which costs more than Standard would have.
Both errors come from the same place: deciding where something goes without knowing how often it moves.
That's not a cloud skill. That's inventory classification, and it's been around for decades.
Where the analogy breaks
Here's where I have to dismantle my own comparison, because it stops working in an important place.
In a physical warehouse, space is finite. That constraint is what forces the whole discipline. If I don't move the slow stock to the back, I physically run out of room, and the problem announces itself.
In S3, space is effectively unlimited. Nothing runs out. Nothing jams. No one trips over a pallet in the aisle.
Which means the discipline has to come from somewhere else entirely — from lifecycle policies you configure deliberately, and from actually looking at your bill. The warehouse punished bad decisions immediately and physically. The cloud punishes them quietly, on a monthly invoice, for years.
I honestly think that's the harder version of the problem.
If you came to cloud from an operational background — logistics, production, maintenance — I'd be curious which concept clicked fastest for you because you'd already lived it somewhere else. For me it was this one, and it was almost unfair how quickly it landed.
Reference: Amazon S3 storage classes documentation.
Enjoyed reading this content? Let the author know!
Your likes, comments, shares, and saves help creators reach more builders.
Loading recommendations
Loading article