Are snapshot and backup fees quietly inflating monthly cloud spend?
Many organizations see storage, API, and egress charges climb as retention windows extend.
This can produce surprise restore and rehydration bills.
For digital entrepreneurs, sysadmins, and small IT teams handling budgeting and reliability, provider-level line-item comparisons help.
They show which retention policies cut costs and which raise risk-adjusted TCO.
Comparativa rápida, provider snapshot TCO
The table below shows modeled monthly totals for two representative workloads in US regions.
It also lists the main secondary charges to include when you budget.
The assumptions used sit after the table.
| Provider |
Region |
Storage $/GB-mo (2024) |
API / indexing est. |
Egress $/GB |
Example: 100×100GB VMs (10 TB) |
Example: 10 TB DB (1% daily change) |
| AWS EBS (snapshots) |
US East (N. Virginia) |
$0.05/GB-mo |
$20–$150/mo (varies with ops) |
$0.09/GB cross-region |
7d: $535 · 30d: $650 · 90d: $950 · 365d: $2,325 |
7d: $215 · 30d: $400 · 90d: $800 · 365d: $2,100 |
| Microsoft Azure (managed disks) |
East US |
$0.06/GB-mo |
$30–$200/mo |
$0.087/GB cross-region |
7d: $642 · 30d: $780 · 90d: $1,140 · 365d: $2,790 |
7d: $258 · 30d: $480 · 90d: $960 · 365d: $2,520 |
| Google Cloud (PD snapshots) |
us-central1 |
$0.026/GB-mo |
$15–$120/mo |
$0.12/GB cross-region |
7d: $279 · 30d: $340 · 90d: $540 · 365d: $1,320 |
7d: $112 · 30d: $210 · 90d: $500 · 365d: $1,260 |
| DigitalOcean |
NY3 |
$0.05/GB-mo |
$10–$50/mo |
$0.01–$0.09/GB |
7d: $535 · 30d: $650 · 90d: $950 · 365d: $2,325 |
7d: $215 · 30d: $400 · 90d: $800 · 365d: $2,100 |
| Backblaze B2 (object backup) |
US |
$0.005/GB-mo |
$5–$40/mo tooling ops |
$0.01/GB eg. Download |
7d: $60 · 30d: $75 · 90d: $115 · 365d: $285 |
7d: $24 · 30d: $60 · 90d: $125 · 365d: $260 |
Assumptions used: US region pricing (2024).
Workload A equals 100 VMs × 100 GB with one baseline full snapshot.
Daily changed data equals 1% of each disk.
Workload B equals a 10 TB DB with 1% net daily change.
Storage effective size equals baseline plus the sum of daily deltas retained.
API/indexing and egress are estimated for typical enterprise patterns and shown as ranges.
7-day cost
AWS $535
GCP $279
Backblaze $60
A clear line-item worked example removes the guesswork when reconciling table totals with vendor bills.
- For example, use a 10 TB baseline with 30 days retention and 1% daily change.
- Assume storage price $0.05/GB-mo, egress $0.09/GB, and API/indexing $100/month.
- Storage math: (10 TB + 30 × 0.1 TB) = 13 TB.
- 13,000 GB × $0.05 = $650 storage.
- API/indexing = $100.
- Assume one cross-region restore per month of a 1 TB snapshot.
- 1,000 GB × $0.09 = $90 rehydration/egress.
- Total modeled monthly snapshot cost = $650 + $100 + $90 = $840.
Include a worked line-item like this alongside each provider row.
That makes explicit which part of the total is storage, API, or restore/egress.
Teams can swap in measured API counts or restore frequency to get accurate TCO.
AWS
AWS often fits teams that need tight integration with EC2 and broad regional coverage.
The platform offers mature snapshot tooling and multiple lifecycle options.
Secondary fees can change decisions quickly.
Pros
AWS snapshots integrate with AMIs and offer incremental storage accounting.
Many automation tools and community modules exist for snapshot lifecycle management.
Cons
The most frequent error at this point is assuming snapshot cost equals VM disk size.
API operations, cross-region copy egress, and snapshot indexing can add hundreds to thousands per month.
That leads to surprise bills when teams copy snapshots for DR.
For whom it fits
Choose AWS if you require multi-AZ or multi-region backups and fast restores within AWS.
Avoid AWS if egress-heavy migrations dominate your budget.
The next section shows cloud differences clearly.
Microsoft Azure
Azure gives tight integration with managed disks and backup vaults.
It can simplify compliance for many enterprise customers.
Pricing depends on snapshot types and redundancy choices, which affect totals.
Pros
Azure snapshot policies and Backup Vault offer GUI-driven lifecycle rules and soft-delete features.
Native integration with Azure Policy simplifies enforcement.
Cons
This works well in theory, but in practice Azure snapshot storage and geo-redundant options can double or triple price per GB.
Teams that compare headline $/GB without normalizing redundancy get incorrect rankings.
For whom it fits
Choose Azure for Windows-heavy stacks and Microsoft licensing needs.
Choose it when policy management and enterprise compliance matter more than raw storage price.
A short summary follows.
Google Cloud
GCP snapshots often show the lowest $/GB for standard snapshots in many US regions.
They have simple incremental behavior and can win on long retention when restore patterns are predictable.
Pros
Lower snapshot storage rates in many US regions reduce long-term retention cost.
GCP offers regional snapshots and fast restores.
Cons
Be careful with cross-region replication and egress.
Rehydrating or migrating large volumes can incur significant transfer fees.
Teams that expect frequent full restores must model egress explicitly.
For whom it fits
Choose GCP when storage price sensitivity is high and RTO targets allow measured restores.
Choose it when the team can use Google's snapshot tooling and monitoring.
Other providers: DigitalOcean and Backblaze
Some providers trade features for price.
DigitalOcean is easy to operate for small fleets.
Backblaze B2 offers the cheapest raw object storage and can lower long-term archive costs dramatically.
Pros
Backblaze gives low $0.005/GB-mo storage (2024), which suits long-term cold archives.
DigitalOcean combines simple console access with predictable pricing for small-to-medium workloads.
Cons
Using object storage like Backblaze requires external tools such as Restic, Borg, or Veeam.
That adds operational complexity and can increase restore times.
DigitalOcean lacks the deep lifecycle features of hyperscalers.
For whom they fit
Choose Backblaze for long, cold retention where restores are rare and RTO can be hours.
Choose DigitalOcean for simpler, developer-driven stacks with modest retention needs.
How to choose according to your situation
Map RTO and RPO to recovery windows, then run a TCO scenario that includes storage, API, egress, and restore costs.
Short windows often cost more per GB because per-snapshot overhead is proportionally higher.
Decision steps
- Define RTO and RPO for each workload class.
- Measure changed-blocks per day and compression or dedupe potential.
- Plug numbers into a TCO model that includes API, indexing, egress, and restore fees.
Simple rule of thumb
If RTO and RPO are one hour or less, choose daily snapshots retained seven to thirty days on the same provider.
If RTO can be several hours and retention is long, add archive to a cheap object store and keep a weekly snapshot on the primary cloud.
Example profiles
- Critical DB (RTO 30m, RPO 15m): daily snapshots plus PIT logs retained 365 days in primary cloud.
- Dev VMs (RTO 4h, RPO 24h): seven-day retention; a cheaper provider is acceptable.
To turn RTO and RPO into a budget decision, quantify the dollar cost of improving RTO and compare to business benefit. For example, compare two retention designs for a 10 TB database. Option A: hot daily snapshots retained 30 days on the primary cloud yield effective stored size 13 TB and storage cost $650 per month. Option B: weekly hot snapshot plus daily increments archived to Backblaze at $0.005/GB-mo keeps a 10 TB hot copy for $500 per month.
Archived deltas for 30 days equal about 3 TB and cost about $15 per month. Total for Option B is roughly $520 per month, but full restores need hours to rehydrate.
If the business values reducing worst-case downtime by three hours, the incremental cost of Option A over B is roughly $130 per month.
That equals about $43 per reduced downtime hour.
Present several quantified tradeoffs like dollars per hour of RTO improvement to help teams pick retention that balances cost and recovery objectives.
What nobody tells you about snapshot bills
Many teams treat snapshots as passive storage and ignore hidden drivers that dominate bills as retention grows.
The real cost combines retained unique blocks, request volume, and restores.
Some providers charge per-snapshot metadata indexing or count operations that add non-storage line items.
Those charges add up fast for fleets with thousands of snapshots.
Chaining and effective retained TB
A common case: 100 VMs with 1% daily change produced a 46 TB effective retained size after one year.
That example came from an audit years ago and pushed a bill from hundreds to thousands per month.
Deduplication and compression
Dedupe and compression behavior varies by provider.
Testing on real data shows effective savings ranging from 10% to 70% depending on workload type (2023–2024 internal tests and vendor docs).
Do not assume vendor claims map to a given dataset.
Dedup and compression behavior changes effective retained TB and must be measured per workload and provider.
A practical measurement: take a representative 100 GB sample, capture a full snapshot, then record 7–14 incremental snapshots.
Record the billed storage size reported by the provider or object storage client.
Across many workloads, savings vary widely: VM images, log consolidation, or database pages often show large reductions.
Compressed or encrypted media often show under 10% benefit.
Translate the observed dedupe or compression factor into your TCO model.
For example, a 30% dedupe factor reduces a 46.5 TB effective retained size to about 32.6 TB.
Include provider-specific notes because hyperscalers dedupe at block level while object stores dedupe at object level with client tools.
Practical TCO calculator
The CSV below is copy-paste ready; adjust the parameters in the header and recalc.
Columns: parameter, value, unit.
Use this to build spreadsheets or import into your cost tool.
Parameter,value,unit
workload,100 VMs x 100 GB,total GB
baseline_size,10000,GB
daily_change_pct,1,%
days_retained,30,days
storage_price,$0.05,USD per GB per month
egress_price,$0.09,USD per GB
api_index_estimate,$100,USD per month
Automation snippets and retention templates
Automation reduces human error and prevents orphaned snapshots that accumulate costs.
The examples below are minimal and need IAM and rate-limit adjustments.
AWS CLI: safe prune
bash
aws ec2 describe-snapshots --owner-ids self --filters "Name=tag:Backup,Values=Daily" /
--query 'Snapshots[?StartTime<2023-01-01].{Id:SnapshotId,Time:StartTime}'
hcl
resource "aws_dlm_lifecycle_policy" "daily" {
description = "Daily snapshots, retain 14"
policy_details { / minimal / }
}
Retention templates
- Critical DBs: daily snapshots (retain 30 days), weekly full export to archive (retain 365 days), PIT logs retained 90 days.
- App servers: daily snapshot retained 7 days, weekly retained 30 days.
Use provider snapshots for fast local restores, but move cold copies to cheap object storage when RTO allows it.
This tradeoff can save up to 60% on long-term retention but lengthens restore time to hours.
Test restores quarterly and keep one tested restore candidate on the primary cloud to meet short RTOs.
Estimated example: For a 10 TB active dataset with 1% daily change, keeping 365 daily snapshots on AWS at $0.05/GB-mo yields an effective snapshot size ~46.5 TB and storage cost ~$2,325/mo (2024 pricing and the model above). Use measurement-first approach to avoid surprises.
Sources and further reading
The pricing used above is based on public vendor pages and common enterprise measurements (2024).
For vendor pricing details see AWS EBS pricing and consult the NIST SP 800-53 guidance for retention and access controls: NIST SP 800-53.
Warning: Do not rely on headline $/GB without normalizing for redundancy (LRS vs GRS, multi-region) and restore fees. That error causes most budget misses.
If a tailored TCO run for your exact profile would help, copy the CSV above, insert your changed-block measurements, and run it in a spreadsheet; a short review of the resulting numbers will show whether archive plus a small hot copy is the cheapest safe path for your RTO/RPO.
Frequently asked questions
What drives snapshot costs beyond storage?
API calls, snapshot indexing, cross-region copies, and restore/rehydration fees often add more to monthly bills than raw GB storage. Account for request volume and any cross-region transfer in your model.
How to measure effective retained TB for my workload?
Capture an initial full snapshot, then collect N incremental snapshots and record changed GB per snapshot. Average the delta and multiply by the number of days retained, then add baseline unique blocks to estimate effective retained TB.
Are archive tiers always cheaper for long-term retention?
Archive storage is cheaper by $/GB but adds rehydration fees and multi-hour restore times. Use archive when restores are infrequent and RTO can tolerate the delay.
How often should I run restore drills?
Run a restore drill at least quarterly for mission-critical systems and semi-annually for less-critical systems. The drill should validate both integrity and the actual RTO under realistic conditions.
How to account for compliance retention requirements?
Map legal retention minimums to specific lifecycle rules and include immutability/WORM premiums. Some regulations require geographic residency; factor multi-region replication and audit costs into the TCO.
Is object storage plus Restic/Borg a viable option?
Yes for long-term archive and cost reduction, provided the team accepts longer restore times and maintains restore automation. Measure dedupe and compression on sample data before deciding.
Which providers hide snapshot and egress fees?
Hyperscalers commonly disclose fees but present $/GB in a way that omits request and cross-region charges in quick comparisons. Always check the detailed pricing pages and include cross-region egress and API costs in TCO.
