About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.
Three measured figures decide the storage part of a backup bill. They are how much data is protected, how much of it is new or changed each day, and how long restore points are kept. Add the provider's charge structure and you can build a budget with a low, expected and high case. You do not need to understand the technology to do that. You need IT to give you three numbers and the provider to tell you which one each charge is based on.
The figures here are measured ones, and the charges are described as the providers describe their own. Tax and accounting treatment is a matter for your accountant and is not covered.
Which three figures should IT supply?
| Figure | Plain meaning | Where IT finds it | Moves when |
|---|---|---|---|
| Protected size | Used data on the servers being backed up (not the disk size) | Server disk usage, or the provider's report | The business adds data |
| New data per day (change rate) | How much data is new or changed each day | The backup report's per-run "added" figure, or a percentage | Databases, mail, logs and exports grow or are rewritten |
| Retention | How many daily, weekly, monthly and yearly points are kept | The written backup policy | Policy or compliance needs change |
Microsoft's own cost estimator for Azure Backup asks for exactly these. It wants "the used size of disks or servers", the expected "churn" ("the amount of change in data"), and how long to keep daily, weekly, monthly and yearly backups. Its example: "if you had a VM with 200 GB of data to be backed up and 10 GB of it changes every day, the daily churn is 5%". It adds that "higher churn will mean that you back up more data".
The change rate matters more than people expect, because of how stored size builds up. The first backup stores the protected size once. Each later point adds only the new data. In a store that keeps each block once, restic's manual describes the effect: "for modified files, only modified Blobs have to be saved in a subsequent backup".
How does a provider turn those figures into a charge?
Providers split the bill in different ways. Two published structures show the range. The wording below is each provider's own, with no prices.
| Structure | What the charge is based on | What to ask |
|---|---|---|
| Azure Backup (Microsoft pricing page) | A protected-instance charge in size bands, "plus storage consumed". Size is measured "before compression and encryption" | Is the band based on protected size? How do bands step at your size? |
| Azure Backup storage | A per-GB charge by redundancy choice (locally, zone or geo-redundant), separate from the instance charge | Which redundancy is in the quote? |
| AWS Backup (AWS pricing page) | Storage "billed in a month is based on the average storage space used throughout the month (billed as GB-Month)"; restore billed on "the amount of data restored for the month"; data copied between Regions billed | How is the average taken? What is restore billed on? |
Two details change a budget. First, Microsoft says an instance over 500 GB is billed "for each 500 GB increment", with a worked example of 1.2 TB charged as two 500 GB increments plus the remaining 200 GB. So that part of the bill steps as your data grows, instead of rising smoothly. Second, AWS shows how a month average works: 400 GB for 15 days and 800 GB for 15 days "= 18,000 GB-Days", divided by 30 "= 600 GB-Month". Growth during the year is therefore billed gradually, not all at once.
Some resources are stored in full each time. AWS says "not all resource types support incremental backups. For resources that do not support incremental backups, each backup is a full copy, which can result in higher storage costs". A quote that does not say which applies cannot be budgeted.
Which lines are fixed and which vary?
| Line | Behaviour | Budget treatment |
|---|---|---|
| Protected-instance or per-machine charge | Steps with size or count of machines | Fixed between bands; recalculate when data crosses a band |
| Stored data (GB-month) | Rises with protected size, change rate and retention | Variable; use a range |
| Redundancy choice | A rate per GB that depends on the copy options chosen | Fixed once chosen |
| Restores | Per GB restored, or free for some tiers | Variable; budget a small amount, and a larger one for a recovery event |
| Copies between regions or out of the provider | Charged per GB moved | Variable; ask for the rate |
| Archive tiers | Minimum storage periods and retrieval fees | Ask for the minimum period and the retrieval charge |
Microsoft's pricing page notes that restoring from its Standard tier "is free of charge and does not incur any transaction or egress charges", while Archive tier data has a "one-time data retrieval fee" and an early deletion period of 180 days. AWS lists a 90-day minimum for its cold storage. Ask for the minimum period if cold or archive storage is quoted.
How do I build a budget with a range?
Use this formula for the storage part, with the three figures from IT and a rate R per GB-month taken from the quote:
stored size at a point in time = protected size + (daily new data x days of retention)
monthly storage cost = average stored size x R
This assumes the daily change is new to the store each day and is kept for the full retention. Where changes overlap or the same block is rewritten, the real figure is lower. The formula is an estimate for budgeting. It is not what the provider will invoice.
Example (illustrative): fs01 holds 2,000 GB of used data. Data grows 10% a year. About 1% of it is new each day, kept for 30 daily points. The oldest point holds the full copy, so 29 days of new data sit on top of it. R is a placeholder rate per GB-month.
- Start of year: 2,000 GB + (20 GB x 29) = 2,580 GB.
- End of year: 2,200 GB + (22 GB x 29) = 2,838 GB.
- Average over the year, assuming steady growth: (2,580 + 2,838) / 2 = 2,709 GB.
- Annual storage: 2,709 GB x 12 months = 32,508 GB-months, so 32,508 x R.
- High case, 2% daily change: start 2,000 + (40 x 29) = 3,160 GB; end 2,200 + (44 x 29) = 3,476 GB; average 3,318 GB; 39,816 GB-months.
- Low case, 0.5% daily change: start 2,000 + (10 x 29) = 2,290 GB; end 2,200 + (11 x 29) = 2,519 GB; average 2,404.5 GB; 28,854 GB-months.
The range is 28,854 to 39,816 GB-months, and the high case is about 38% above the low case. The change rate alone produced that spread, which is why IT's measured figure matters more than any rule of thumb.
How do I check the figures with IT and the provider?
Ask IT:
- What is the used size of each protected server today, and a year ago?
- How much new data did the last 30 backups add, per server? (restic prints an "Added" line for each run; ask what your tool's equivalent is.)
- What is the written retention policy, and does the tool apply it? A policy on paper that is not applied is the usual cause of a bill above budget.
- Which servers hold compressed or encrypted archives, database exports or video? These change in full when edited.
Ask the provider:
- Is the charge based on protected size or stored size, and is that figure before or after compression and encryption?
- If the charge is stored size, is it a monthly average or the end-of-month figure?
- Is each backup stored in full, or is only new data stored, for each type of server we back up?
- Which charges apply when we restore, copy to another location or stop protecting a server? Does billing continue for the data while it is kept?
- What are the minimum storage periods and early-deletion terms?
A quote that cannot answer questions 1 and 3 cannot be compared with another. The comparison guide for managed hosting quotes sets out how to line quotes up on a worksheet.
Read next
Related articles you may want next.
- HostingForecasting capacity from a year of measurements: when is the next upgrade due?How a business turns twelve months of storage, memory and processor figures into a likely date and budget line for its next hosted-server upgrade, using two simple forecasting methods with worked arithmetic.Read article
- Backup & RecoverySize Backup Retention Using Measured Storage GrowthBuild a capacity forecast from stored growth, copy formats and retention dependencies, with explicit headroom, recovery needs and uncertainty.Read article
- Backup & RecoveryWill keeping backups twice as long double the storage cost?Doubling backup retention can double storage or add far less. Worked sums for full copies, new-data-only storage and weekly fulls, plus questions to ask.Read article
Risks and limits
- Risk: cutting retention to save money removes older restore points and cannot be reversed once they are gone. Make that decision against the recovery need, not the invoice.
- The formula assumes new data each day is unique and kept for the whole retention. It ignores compression, metadata overhead and block rounding, which vary by product.
- The Azure and AWS pages are examples of structure. Other providers use different units; check each quote against its own terms. No vendor is ranked here.
- Pricing pages change. Re-read the page behind a quote before you sign.
- A budget built from one month of measurements can miss seasonal peaks such as year-end processing. Use at least a full year where you have it; the forecasting guide in the related list shows how.
NetLeverage includes a daily full disaster-recovery backup of each hosted server at no extra charge. Talk to an engineer.
Sources
- Estimate Azure Backup costs and understand pricing, Microsoft Learn, updated 2026-08-26, accessed 17 November 2026
- Azure Backup pricing, Microsoft Azure, date not shown, accessed 17 November 2026 (structure only; no prices quoted)
- Azure VM backup architecture and components (backup costs), Microsoft Learn, updated 2026-01-08, accessed 17 November 2026
- AWS Backup pricing, AWS, date not shown, accessed 17 November 2026 (structure only; no prices quoted)
- What is AWS Backup? (incremental backups), AWS, date not shown, accessed 17 November 2026
- restic: Backing up, restic project documentation 0.19.1, date not shown, accessed 17 November 2026
- restic: Design (backups and deduplication), restic project documentation 0.19.1, date not shown, accessed 17 November 2026
This article was originally published on .