Why Cheap Cloud Storage Is Not Cheap

The storage price is rarely the whole bill

Three other charges do most of the damage:

• Egress — what you pay for data leaving the provider's network.
• Retrieval — what cold and archive tiers charge to make data readable again.
• Per-request API fees — trivial individually, significant at volume.

A comparison that weighs only the per-terabyte storage rate will pick the wrong tier more often than not.

Egress is the one nobody counted

Here's the ratio worth internalising:

Moving 1 TB out costs about what storing that 1 TB for four months costs.

On the illustrative rates in the tool, that's 3.9 months. Which means: if your data leaves regularly, the bill is driven by movement, not volume — and the storage rate you compared providers on turns out to be close to irrelevant.

Watch the egress share of a standard-tier bill as downloads grow, with 5 TB stored:

At the top of that table you are barely paying for storage at all.

Cold tiers have a crossover, and you can calculate it

Archive storage can be five or six times cheaper to hold — while charging a substantial fee every time you read it back.

That's an excellent trade for a backup nobody touches, and a terrible one for anything actively used. The break-even is:

stored TB × (standard rate − archive rate) ÷ retrieval rate

For 5 TB on the default rates, that's 4.75 TB of downloads a month.

And the tool proves it on itself — set downloads to exactly 4.75 and both tiers land on $542.50, with the saving reading exactly $0.00:

Below the crossover archive wins. Above it, the retrieval fees have overtaken the storage saving and the "expensive" tier is cheaper.

The trap that runs the other way

Minimum storage durations. Most cold tiers bill a commitment — commonly 30, 90 or 180 days — whether or not you keep the data that long. Delete an archived file after a week and you can still pay the full ninety days.

Which has a counter-intuitive consequence: for data that turns over quickly, a cold tier can cost more than a standard one before retrieval is even counted.

Check the minimum duration before moving anything short-lived into archive.

Two more things that catch people

Retrieval speed is a separate axis from price. Archive tiers commonly offer expedited, standard and bulk retrieval at very different rates, with restore times from minutes to many hours. A twelve-hour bulk restore is fine for a compliance request and useless during an outage. Decide what restore time you need first, then price that tier.

Cross-region traffic is usually egress too. Replication to another region, or traffic to another provider, is commonly charged — it's generally only movement within a region that's free. This catches people building multi-region redundancy, where the replication itself becomes a recurring line item nobody costed.

About those prices

Every rate in the tool is an input, and deliberately so.

Cloud pricing changes, varies by region and differs between providers. A calculator that shipped with one vendor's numbers baked in would be stale within months and confidently wrong — which is worse than being obviously generic.

Replace all four with your provider's current published figures before trusting any output. What the tool contributes is the shape of the comparison. The numbers have to be yours.

The order to do this in

1. Work out your real monthly egress — it's usually underestimated.
2. Check the minimum storage duration against how fast your data turns over.
3. Confirm the restore time you actually need.
4. Then compare storage rates, which is the step most people start with.