Turning a TB-Sized Bucket Into a GB-Month Line Item
Your object-storage console reports the bucket in terabytes, because that is the scale a real dataset lives at. The rate card you are billed against is written per gigabyte. Somebody has to bridge those two units before the finance meeting, and getting it wrong by a factor of a thousand is the kind of mistake that survives three quarters of planning.
GB = TB × 1000. A 4.5 TB bucket is 4 500 GB. At a representative hot-tier rate of $0.023 per GB-month that is 4 500 × 0.023 = $103.50 for the month — before a single request or byte of egress is counted.What the Invoice Is Actually Measuring
GB-Month, Not GB
One Dataset, Several Rates
Rate Bands Move at TB Boundaries
Both units here are decimal SI, so the multiply-by-a-thousand is exact. The awkwardness on a cloud bill never comes from the conversion; it comes from what the provider decides to count.
Costing a Dataset Before the Invoice Arrives
The workflow below takes a console figure and turns it into a number you can defend in a budget review.
Take the Total From the Usage Report
Pull the average stored size for the billing period, not today's snapshot, and type it into the left field. A comma works as well as a dot if that is how your locale writes 7,5 TB.
Multiply the GB Result by the Tier Rate
The gigabyte figure appears as you type, and that number is your billable GB-months. Multiply it by the published per-GB-month rate for the class to get the storage line directly.
Copy the Figure Into the Model
The copy button on either field puts the bare number on the clipboard with no unit attached — what a spreadsheet cell or a cost-model script wants. Ctrl+C in a field does the same.
Run It Backwards From a Budget Ceiling
Both fields are editable and the swap button reverses direction outright. Divide your monthly allowance by the per-GB rate, type that GB figure into the gigabyte side, and read off how many terabytes the budget buys.
What 8 TB Costs in Each Storage Class
The table takes one stored dataset — 8 TB, or 8 000 GB — across the tier ladder most object-storage providers publish, at representative US-region list rates. Rates vary by provider and region and are revised over time, so treat the dollar column as a shape rather than a quote; the GB-month column never changes.
| Storage class | Billable GB-month | Rate per GB-month | Storage line | Minimum duration |
|---|---|---|---|---|
| Standard / hot | 8 000 GB | $0.023 | $184.00 | None |
| Infrequent access | 8 000 GB | $0.0125 | $100.00 | 30 days |
| Archive, instant retrieval | 8 000 GB | $0.0040 | $32.00 | 90 days |
| Archive, flexible retrieval | 8 000 GB | $0.0036 | $28.80 | 90 days |
| Deep archive | 8 000 GB | $0.00099 | $7.92 | 180 days |
Top to bottom that ladder spans more than twenty to one on the same 8 000 GB, which is why the minimum-duration column matters as much as the rate: a tier at a fifth of the price that bills a mandatory 90 days is only cheaper if the data really will sit still that long.
Every Unit the Rate Card Might Use
The searchable dropdowns on both sides list the full data-storage set, so when a line item is quoted in MB or a quota ticket comes back in TiB you can price it from the same page.
Console Figures Pasted As-Is
Spaces are ignored and a comma is accepted in place of a decimal point, so a value lifted straight out of a usage export converts without reformatting.
Readable at Forecast Scale
Results carry up to eight decimals with thousands separated by spaces, and switch to scientific notation once a figure passes ten digits — useful when a multi-year projection runs off into byte counts.
Both Directions on One Page
Type in either field and the other follows, so the page serves the "how many GB is this bucket" question and the "how many TB fits in this budget" question equally well.
Cloud Storage Billing Questions
Why does my bill show GB-months when I only ever uploaded terabytes?
Providers meter storage continuously and settle in the smallest practical unit, which keeps one rate card usable for customers storing 40 GB and customers storing 40 TB. A gigabyte-month is one gigabyte held for one month, so a 4.5 TB bucket that sat still bills as 4 500 GB-months. If it grew mid-month, the billed figure lands between the starting and ending totals rather than at the peak.
What does a minimum storage duration charge actually add to my bill?
Colder classes commit you to a floor — commonly 30 days for infrequent access, 90 for archive tiers, 180 for deep archive. Delete or re-tier an object early and you are still charged for the remaining days at that class rate. A 2 TB batch dropped into a 90-day tier and removed after three weeks still bills 2 000 GB for the full quarter, which usually erases the saving that motivated the move.
Why did my request costs come out higher than my storage costs?
Storage scales with bytes, requests scale with object count, and the two are unrelated. Five million small writes at a typical $0.005 per thousand is $25 of API charges whether they total 200 GB or 2 TB. Cold classes also apply a minimum billable object size — often 128 KB — so ten million 40 KB files occupying 400 GB bill as 1 280 GB. Packing them into larger archives fixes both problems.
Do the egress lines and the storage lines use the same per-GB rate?
No, and the gap is wide enough to reshape an architecture. Storage bills a recurring rate per GB kept; transfer out bills a one-off rate per GB that leaves, usually after a small free allowance. At representative rates, keeping 2 TB for a month is about $46, while pulling that same 2 TB out once is roughly $171. Both lines convert from the same TB figure, so always ask which behaviour a per-GB price describes.
Does moving 8 TB to a colder class always cut the bill?
Only if the access pattern cooperates. The transition is charged per object, colder tiers add per-GB retrieval fees on top of request costs, and the duration floor starts the moment the object lands. Weigh the difference — 8 000 GB at $0.023 is $184 a month against $32 in an instant-retrieval archive tier — against the reads you realistically expect. Datasets read a few times a year win easily; weekly reads often do not.
No comments yet. Be the first to comment!