Direct answer: at official list prices on 8 November 2026 (USD, US East/US West, excluding tax and discounts), standard snapshot storage is about 0.05/GB-month on AWS EBS Snapshots, 0.05/GiB-month on Azure managed disk snapshots for Standard HDD (about 0.132 for SSD source disks), and 0.05/GiB-month on GCP standard snapshots. Archive snapshots are about 0.0125/GB-month on AWS (90-day minimum) and 0.019/GiB-month on GCP (90-day minimum); Azure has no same-name archive snapshot tier. For a 1 TiB disk with 5% daily change and 30-day retention, AWS/Azure HDD/GCP standard cost about 125.44, Azure SSD about 331.16, and GCP Archive about 47.67; but AWS Archive is a full copy, so 30 archives cost about 384, more than standard incremental. The conclusion: check incremental semantics and retention first, then the per-GB price.
This guide is for cloud platform, SRE, backup/disaster-recovery and procurement teams. It compares AWS EBS Snapshots, Azure managed disk snapshots and Google Cloud Persistent Disk snapshots: standard/archive prices, incremental billing, 1 TiB TCO at 7/30/90-day retention, restore/cross-region costs and an optimization checklist. Prices come from the AWS EBS pricing page, the Azure Managed Disks pricing page and Azure Retail Prices API, and the Google Compute Engine disk/image pricing page (all accessed 8 November 2026). Official price facts, editorial calculations and exclusions are kept separate. This is not procurement advice.
(1) The snapshot bill depends on unique blocks x retention, not disk size; the first snapshot covers used data, later ones only changes. (2) Archive is cheaper per GB, but AWS Archive is a full copy with a 90-day minimum and retrieval fees; daily archiving can cost more than standard incremental. (3) Deleting an old snapshot does not always reduce the bill: later snapshots may still reference that data. Check the snapshot chain, recycle bin and retention policy first.
Basis: standard/archive, incremental/full and units
All three provide block-storage snapshots, but "incremental" means different things: AWS Standard snapshots store changed blocks incrementally, while Archive is a full copy; Azure ordinary snapshots are billed on used capacity, incremental snapshots bill only changes since the previous snapshot, and Premium SSD v2/Ultra instant-access snapshots bill incremental storage hourly; GCP Standard and Archive are both compressed incremental, with a 1-hour minimum for Standard and 90 days for Archive. At minimum, align the unit (AWS labels GB, Azure/GCP label GiB; 1 TiB = 1,024 GiB is about 2.4% above 1,000 GB), whether compression applies, whether the first snapshot is full, retention, retrieval and cross-region network fees.
| Item | AWS EBS Snapshots | Azure disk snapshots | GCP Persistent Disk snapshots |
|---|---|---|---|
| Standard tier | Standard 0.05/GB-month; changed-block incremental; first snapshot full; restore free | Billed on used capacity; incremental bills changes; Standard HDD 0.05, SSD 0.132/GiB-month (LRS) | Standard 0.05/GiB-month; compressed incremental; 1-hour minimum; no retrieval fee |
| Archive tier | Archive 0.0125/GB-month; full copy; 90-day minimum; retrieval 0.03/GB | No same-name archive snapshot tier; use Backup/Blob Archive or export for long-term retention | Archive 0.019/GiB-month; compressed incremental; 90-day minimum; retrieval 0.019/GiB |
| Cross-region/multi-region | Cross-region copy charges data transfer; destination billed at standard snapshot storage | LRS/ZRS snapshots; cross-region via copy/Backup, traffic billed separately | Multi-region Standard 0.083, Archive 0.031/GiB-month; download/upload 0.08/GiB |
| Deletion/retention | Early Archive deletion charged for remaining days; Recycle Bin at snapshot rate | Billed until deleted; later snapshots may still reference old blocks | Archive 90-day minimum; deletion may move data to the next snapshot; Recycle Bin billed |
Sources: AWS EBS pricing page, Azure Managed Disks pricing page, Google Compute Engine disk/image pricing page (accessed 8 November 2026). Editorial summary; Azure full vs incremental snapshots and creation method can change the bill.
Unit prices: standard and archive tiers
Standard-tier list prices are close: AWS 0.05/GB-month, Azure Standard HDD 0.05/GiB-month and GCP 0.05/GiB-month; Azure SSD source disks are 0.132/GiB-month (LRS) and 0.198 for ZRS. Archive differs more: AWS 0.0125/GB-month and GCP 0.019/GiB-month, both with a 90-day minimum; Azure has no same-name archive snapshot, so long-term retention usually uses Backup or object-storage archive. Also account for restore/retrieval: AWS Archive 0.03/GB, GCP Archive 0.019/GiB (0.031 multi-region), and AWS Fast Snapshot Restore 0.75/DSU-hour per snapshot per AZ.
| Item | Price (USD) | Notes |
|---|---|---|
| AWS EBS Standard | 0.05/GB-month | Changed-block incremental; first snapshot covers used blocks |
| AWS EBS Archive | 0.0125/GB-month | Full copy; 90-day minimum; retrieval 0.03/GB |
| AWS Fast Snapshot Restore | 0.75/DSU-hour | Per snapshot, per AZ; per-minute billing, one-hour minimum |
| Azure Standard HDD snapshots | 0.05/GiB-month | LRS/ZRS same price; used capacity/incremental |
| Azure SSD snapshots (LRS) | 0.132/GiB-month | Standard SSD / Premium SSD source disks |
| Azure SSD snapshots (ZRS) | 0.198/GiB-month | Zone-redundant snapshots |
| GCP Standard snapshots | 0.05/GiB-month | Compressed incremental; 1-hour minimum |
| GCP Archive snapshots | 0.019/GiB-month | Compressed incremental; 90-day minimum; retrieval 0.019/GiB |
| GCP multi-region Standard | 0.083/GiB-month | Multi-region replication; Archive 0.031 |
| GCP multi-region download/upload | 0.08/GiB | Multi-region snapshot download/upload |
Sources: AWS EBS pricing page (Standard/Archive, 90-day minimum, retrieval 0.03/GB, FSR 0.75/DSU-hour); Azure Retail Prices API (eastus, 8 November 2026); Google Compute Engine disk/image pricing page (us-central1, 8 November 2026). Public list prices, excluding tax, discounts, commitments and enterprise agreements.
TCO: 1 TiB disk, 5% daily change, 7/30/90-day retention
Assuming the first snapshot is 1,024 GiB, 5% new data per day (51.2 GiB), and steady-state retention, 7 days stores about 1,331.2 GiB, 30 days about 2,508.8 GiB and 90 days about 5,580.8 GiB. At standard unit prices: AWS/Azure HDD/GCP about 66.56 / 125.44 / 279.04; Azure SSD about 175.72 / 331.16 / 736.67; GCP Archive about 25.29 / 47.67 / 106.04. Because AWS Archive is a full copy, 7/30/90 archives cost about 89.60 / 384.00 / 1,152.00, already above standard incremental at 30 days.
| Option | 7-day retention | 30-day retention | 90-day retention |
|---|---|---|---|
| AWS Standard | 66.56 | 125.44 | 279.04 |
| AWS Archive (full copy) | 89.60 | 384.00 | 1,152.00 |
| Azure HDD LRS | 66.56 | 125.44 | 279.04 |
| Azure SSD LRS | 175.72 | 331.16 | 736.67 |
| Azure SSD ZRS | 263.58 | 496.74 | 1,105.00 |
| GCP Standard (uncompressed estimate) | 66.56 | 125.44 | 279.04 |
| GCP Archive | 25.29 | 47.67 | 106.04 |
| GCP multi-region Standard | 110.49 | 208.23 | 463.21 |
Editorial calculation: 1 TiB = 1,024 GiB; first snapshot 1,024 GiB; daily change 51.2 GiB; N-day steady state ~= 1,024 + (N-1) x 51.2 GiB; unit prices as above. Excludes compression (GCP may be lower), data movement after deletion, chain references, recycle bin, network, retrieval, tax and discounts. AWS Archive assumes one full 1,024 GiB copy per day; archiving only weekly/monthly snapshots reduces the count.
Sensitivity: change rate and retention matter more than unit price
For the same 1 TiB disk, raising daily change from 1% to 20% increases 30-day stored data from about 1,320.96 GiB to 6,963.2 GiB and, at 0.05, the cost from about 66.05 to 348.16; at 90 days it rises from about 1,935.36 GiB (96.77) to 19,251.2 GiB (962.56). Ask "how much changes per day and how long must we keep it" before choosing a vendor; moving retention from 7 to 90 days can more than quadruple the bill.
| Daily change | 30-day storage (GiB) | 30-day cost (0.05) | 90-day storage (GiB) | 90-day cost (0.05) |
|---|---|---|---|---|
| 1% | 1,320.96 | 66.05 | 1,935.36 | 96.77 |
| 5% | 2,508.80 | 125.44 | 5,580.80 | 279.04 |
| 20% | 6,963.20 | 348.16 | 19,251.20 | 962.56 |
Editorial calculation: first snapshot 1,024 GiB; daily change = 1,024 x rate; 30 days = 1,024 + 29 x daily; 90 days = 1,024 + 89 x daily; unit price 0.05/GiB-month. Excludes compression, data movement after deletion, chain references, retrieval/network, tax and discounts.
Restore, retrieval and cross-region copy: the costs behind cheap storage
Standard restores usually have no retrieval fee, but the restored disk/volume is billed at normal storage rates; archive restores charge retrieval, and temporary restore may also be billed at the standard rate. Cross-region copy charges data transfer and the destination's standard snapshot storage; GCP multi-region Standard/Archive are 0.083/0.031 per GiB-month, plus 0.08/GiB download/upload. AWS FSR and Azure/GCP instant-access snapshots charge for recovery speed, not capacity, usually hourly or per operation.
| Operation | AWS | Azure | GCP |
|---|---|---|---|
| Standard restore | Free; restored volume billed as EBS | Create a disk; disk billed | No retrieval fee; cross-region network charges |
| Archive restore | 0.03/GB retrieval + temporary standard storage | No archive snapshot tier | 0.019/GiB; multi-region 0.031/GiB |
| Cross-region copy | Data transfer + destination storage | ZRS snapshot or copy/Backup + traffic | Multi-region Standard 0.083; Archive 0.031 |
| Fast restore | FSR 0.75/DSU-hour/snapshot/AZ | Instant access hourly + restore capacity fee | Instant snapshot 0.05/operation + differential storage |
| Early deletion | Archive charged for remaining days | Billed until deleted; chain references may persist | Archive 90-day minimum; chain references/recycle bin |
Sources: AWS EBS pricing page; Azure Managed Disks pricing page; Google Compute Engine disk/image pricing page (accessed 8 November 2026). Editorial summary; restore speed, capacity charges, network paths and minimum billing periods follow the official pages and bill.
The snapshot chain: why deleting old snapshots may not cut the bill
Incremental snapshots form a chain: later snapshots may reference blocks from earlier ones. Deleting a middle snapshot does not necessarily free the blocks it references; GCP states that deleting a full or incremental snapshot may move some data to the next incremental snapshot, increasing later storage cost. AWS and Azure have similar block-reference behavior. The order should be: stop unneeded retention policies, delete from the oldest/most independent end, then consider middle snapshots; verify with storage metrics and the bill, not the snapshot count.
| Behavior | AWS | Azure | GCP |
|---|---|---|---|
| Incremental unit | Changed blocks | Changed blocks/used capacity (snapshot type dependent) | Compressed changed blocks |
| First snapshot | Full used blocks | Used capacity | Compressed used data |
| Delete a middle snapshot | Referenced blocks still billed | Same | Data may move to the next snapshot |
| Recycle bin/soft delete | Recycle Bin at snapshot rate | Soft delete billed separately | Recycle Bin at standard rate |
| Minimum billing | Archive 90 days | Actual existence time | Standard 1 hour; Archive 90 days |
Sources: AWS EBS pricing page (Archive 90 days, Recycle Bin); Azure Managed Disks pricing page; Google Compute Engine disk/image pricing page (Standard 1 hour, Archive 90 days, data movement after deletion). Editorial summary; chain references and deletion behavior follow vendor documentation and the actual bill.
How to choose: RPO, retention, restore speed and compliance
Start with four questions: RPO/RTO (minutes or hours), retention, cross-region/immutability, and restore frequency. Use standard incremental for frequent short retention; use archive for long-term, rarely restored data, but check whether it is a full copy or compressed incremental; buy FSR/instant access only when minutes-level restore is required; snapshots in the same account and region are not an offsite backup. If the requirement is application consistency rather than block-level recovery, prefer a backup service (AWS Backup/Azure Backup) over raw snapshots.
| Scenario | Prefer | Check first |
|---|---|---|
| Daily backup, 7-30 day retention | AWS/Azure/GCP standard incremental | Daily change rate, first full, deletion policy |
| Long-term compliance, rare restore | AWS Archive or GCP Archive | Full vs incremental, 90-day minimum, retrieval |
| Minutes-level restore | AWS FSR, Azure/GCP instant access | Hourly/operation fees, AZ/region coverage |
| Cross-region/multi-region DR | Snapshot copy/multi-region snapshots | Transfer, destination storage, RPO/RTO |
| Application consistency | AWS Backup/Azure Backup/vendor agent | VSS/quiesce, database logs, restore drills |
| Ransomware/accidental deletion | Immutable backup/recycle bin/cross-account | Retention lock, delete/restore permissions |
| Dev/test disks | Short-retention standard snapshots | Orphan snapshots, expired images, tagging |
Editorial selection framework, not a vendor recommendation; validate RPO/RTO, compliance retention and restore requirements against enterprise standards and vendor documentation.
90-day cost-reduction checklist
Use the first 90 days to establish which disks are snapshotted, for how long, how much changes daily and who owns them. The order is inventory, retention policy, incremental/archive tiering, orphan cleanup, restore drills, procurement. The most common waste is "90 days for every disk", "archive every day", "deleted snapshots but the chain remains" and "snapshots treated as offsite backup".
| Phase | Action | Output |
|---|---|---|
| Weeks 1-2 | Inventory volumes/disks, snapshot count, size, retention and owner tags | Snapshot list, cost baseline |
| Weeks 2-3 | Tier by RPO: critical/normal/dev with different retention | Retention policy, owners |
| Weeks 3-5 | Use standard incremental for frequent snapshots; archive only long-term, rarely restored data | Tiering plan, archive count |
| Weeks 5-7 | Delete orphan/expired snapshots; check chain references and recycle bin | Deletion list, actual savings |
| Weeks 6-8 | Run a restore drill and record RTO and restore cost | Drill report, RTO |
| Weeks 8-10 | Assess cross-region/immutable/backup services to close DR gaps | DR plan, budget |
| Weeks 10-12 | Set budget alerts, monthly review, automated cleanup | Alerts, automation |
Editorial optimization advice, not an official template; snapshot, backup, compliance and restore requirements follow enterprise standards and vendor documentation.
FAQ
Are snapshots billed by disk size or actual data?
By unique data blocks actually stored, not by provisioned disk size. The first snapshot usually covers all used data, later ones only changed blocks/data; GCP also compresses. Empty blocks and unused capacity are generally not billed, but the vendor's metering is authoritative.
How much does a 1 TiB disk with daily snapshots and 30-day retention cost?
At 5% daily change and a 0.05 standard rate, 30 days stores about 2,508.8 GiB, or about 125.44; Azure SSD at 0.132 is about 331.16; GCP Archive at 0.019 is about 47.67; AWS Archive, being a full copy, is about 384.00. Actual cost depends on change rate, compression, chain references after deletion and retrieval.
Is archive always cheaper?
No. Archive has a lower per-GB price, but AWS Archive is a full copy with a 90-day minimum and retrieval fees; archiving 30 daily copies can cost more than standard incremental. GCP Archive is compressed incremental and usually better for long-term retention, but retrieval and the 90-day minimum still apply.
Why did the bill not drop after deleting old snapshots?
Incremental snapshots are chained; later snapshots may reference old blocks. GCP also states that some data may move to the next snapshot after deletion. Stop retention policies first, delete from the oldest/most independent end, and verify with storage metrics and the bill.
Can snapshots replace backups?
Not on their own. Snapshots in the same account and region can be lost with account compromise, accidental deletion or a regional failure; keep cross-account/cross-region/immutable copies and test restores. Application consistency also needs VSS/quiesce/backup agents; block snapshots are usually crash-consistent.
How are cross-region copy and restore billed?
Cross-region copy charges data transfer and destination standard snapshot storage; GCP multi-region Standard is 0.083 and Archive 0.031 per GiB-month, plus 0.08/GiB download/upload; AWS bills standard snapshot storage in the destination region after copy; Azure can use ZRS snapshots or copy/Backup, with traffic billed separately.
Do snapshots affect production performance?
Routine incremental snapshots are asynchronous in the background and usually have less impact than the first full snapshot; heavy restores, FSR/instant access or cross-region copy can consume resources. The impact depends on volume type, workload and implementation, so test at off-peak rather than assuming "no impact".
How can snapshot costs be optimized?
Tier retention by RPO, shorten unnecessary 90-day policies, use standard incremental instead of daily archive, delete orphan/expired snapshots, check chain references/recycle bin, move long-term rarely restored data to archive, buy FSR/instant access only for minutes-level recovery, and set budget alerts plus monthly reviews.
Research method and limitations
This article separates "official price facts" from "editorial calculations." AWS Standard 0.05/GB-month, Archive 0.0125/GB-month, Archive retrieval 0.03/GB, the 90-day minimum, full-copy Archive, free standard restore and FSR 0.75/DSU-hour come from the AWS EBS pricing page. Azure Standard HDD 0.05, SSD 0.132 (LRS), SSD ZRS 0.198/GiB-month and the snapshot billing description come from the Azure Managed Disks pricing page and Azure Retail Prices API (eastus, 8 November 2026). GCP Standard 0.05, Archive 0.019, multi-region Standard 0.083, Archive 0.031, retrieval 0.019/0.031, download/upload 0.08/GiB, Standard 1-hour/Archive 90-day minimums and compressed incremental come from the Google Compute Engine disk/image pricing page (us-central1, 8 November 2026). TCO, sensitivity, selection framework and the 90-day checklist are editorial.
Eight limitations: (1) only public list prices, excluding tax, enterprise discounts, commitments, support and migration; (2) only US East/US West, other regions differ; (3) capacity normalized to 1 TiB = 1,024 GiB, while AWS GB and Azure/GCP GiB metering can differ; (4) assumes 5% daily change, incremental snapshots and steady-state retention, with no real change-rate/compression test; (5) excludes data movement after deletion, chain references, recycle bin and soft delete; (6) excludes cross-region network, retrieval, FSR/instant access and restore drills; (7) does not compare AWS Backup, Azure Backup, third-party backup or object-storage archive; (8) prices and features change, so recheck the official pages and bill before publishing or purchasing.
Sources and basis
- AWS, Amazon EBS Pricing: aws.amazon.com/ebs/pricing. Standard/Archive, 90-day minimum, full-copy Archive, retrieval 0.03/GB, FSR 0.75/DSU-hour, free standard restore. Accessed 8 November 2026.
- Microsoft Azure, Managed Disks Pricing: azure.microsoft.com/en-us/pricing/details/managed-disks. Standard HDD/SSD, ZRS snapshots, incremental/full snapshot billing. Accessed 8 November 2026.
- Azure Retail Prices API (armRegionName=eastus, meterName contains Snapshot): prices.azure.com/api/retail/prices. Standard HDD 0.05, SSD LRS 0.132, SSD ZRS 0.198, instant-access components. Accessed 8 November 2026.
- Google Cloud, Disk and image pricing: cloud.google.com/compute/disks-image-pricing. Standard 0.05, Archive 0.019, multi-region 0.083/0.031, retrieval/download, 1-hour/90-day minimums, compressed incremental. Accessed 8 November 2026.
Data cut-off 8 November 2026; prices and features follow the official AWS/Azure/Google pages and bills; this is not procurement, commercial or legal advice.