Here's the problem: cloud backup costs are rising fast, but storage rates aren't the reason. The part people miss is that charges hide in three places teams stop looking after setup: snapshot sprawl, versioning multiplication, and archive retrieval penalties.
This keeps happening because cloud environments grow faster than the governance around them. Teams spin up resources, accounts multiply, and no single dashboard shows what's actually being backed up across the organization.
How much do cloud backups actually cost?
At the unit-rate level, cloud backup storage runs from about $0.001 per GB per month for deep archive tiers to roughly $0.023 per GB per month for hot object storage. For a 2TB workload, that is the difference between $2 and $47 per month.
At enterprise scale, around 500TB of structured and unstructured data, the gap widens dramatically once API calls, egress, and retention penalties enter the bill.
Note: Cost estimates require verification based on specific usage patterns.
Here is how the major providers compare on storage pricing alone:
In practice, these unit rates are just the entry point. The real costs come from three failure modes that compound over time, and most teams discover them during billing audits, not daily operations.
Charge 1: Snapshot sprawl in VM and EBS environments
This is the most expensive backup category that most teams don't track. EBS volumes automatically generate snapshots, which accumulate across accounts and regions. After a year of operation, it is normal for an enterprise environment to contain thousands of snapshots that no one can identify.
We have sat with customers who were running cleanup audits in which a single AWS account held more than 1,000 snapshots, none of which the team could match to an active workload.
Multi-account environments compound this fast. Every account that runs production has snapshots, and every account that ran production once still has them, often with no centralized view to surface them. This breaks because cloud environments grow faster than the governance around them. Teams spin up resources, accounts multiply, and no single dashboard shows what's actually being backed up across the organization.
The cost shows up in two places. The first is direct storage charges for snapshots that should have been deleted months ago. The second is recovery charges when teams have to restore an entire 400GB volume to pull back three rows from one database table.
Native cloud snapshots do not index file or record contents. They are block-level copies of the entire volume, which means recovery is all-or-nothing.
We worked through a cost breakdown with a team that needed three database rows back. Their only option was a full 400GB volume restore. The bill came to $36 in egress alone, plus compute time and engineering hours, for a total incident cost of over $200.
Run that a few times per quarter, and it stops being a rounding error.
Cloud Backup Posture Management (CBPM) solves the visibility side. Eon continuously discovers and classifies cloud resources across accounts and regions, so unattached snapshots, orphaned volumes, and over-retained backups surface automatically.
Separate benefit: granular recovery eliminates the restore cost trap entirely. Instead of paying $36 in egress to restore 400GB for three database rows, you pull back exactly what you need.
After switching to Eon, StructuredWeb's recovery time dropped 98%, from hours of digging through unindexed snapshots to 10 to 15 minutes per job.
Charge 2: Versioning and replication growth in object storage
This breaks because ownership is distributed across teams and accounts. S3 and equivalent object stores carry a completely different cost dynamic than VM backups. There are no snapshots to clean up, but versioning and replication compound storage charges quietly in the background.
Enabling S3 versioning is the right call for ransomware protection. It prevents an attacker from overwriting the only copy of a file.
But versioning without a lifecycle policy means that every change to every object creates a new full version, stored at full price. A 1GB file edited daily for a year accumulates 365GB of storage. Across a bucket with millions of objects, this stacks into a very large monthly line item that teams did not budget for. This keeps happening because versioning feels like a safety net until the bill arrives. Teams enable it for ransomware protection, but forget that 'versioning' means 'full new copy every time.
Replication makes the math worse. Cross-region replication for compliance or resilience usually requires versioning to be enabled on both the source and destination.
Old versions on the source replicate forward, the destination accumulates them too, and the total storage footprint can easily run two to three times what teams expected.
Lifecycle policies that expire non-current versions exist on every major provider, but they are often missing or misconfigured. The default keeps everything indefinitely.
The third common gap is forgotten buckets. Production teams spin up buckets for new services, the service gets retired, and the bucket sits there for years collecting tiny daily uploads from a process the team has long since stopped tracking.
We reviewed an Azure bill with more than 5,000 line items, in which three different backup tools were running across the environment, with no unified view. Separating intentional spending from waste took weeks.
CBPM addresses this by mapping every bucket and storage account to a classified workload, applying a retention and lifecycle policy that matches the actual data type, and flagging drift the moment it appears.
Eon's Cost Explorer breaks this out across AWS, Azure, and Google Cloud, so spikes are traceable to the bucket and account they originated in.
Charge 3: Archive tier traps
Glacier and equivalent archive tiers are cheap on paper. In practice, two charges erase most of the savings the moment a recovery job runs.
The first is minimum retention. Every major archive tier bills for a minimum number of days, regardless of when the data is deleted. Glacier Flexible bills for 90 days, Glacier Deep Archive for 180 days, and Google Cloud Archive for 365 days.
If your rotation cycle is shorter than the minimum window, every deleted object still bills for the remaining days. A team running a 30-day backup cycle on Glacier Flexible is paying for 60 extra days on every object they touch.
The charge does not show up as a line item. It blends into the storage total, and by the time someone audits, the overcharge has been running for months.
The second charge is retrieval, both in fees and in time. Archive tiers are designed for data you almost never need, and the pricing reflects that on both ends.
Glacier Deep Archive storage runs $0.00099 per GB per month, but standard retrieval costs $0.02 per GB and takes 12 hours; bulk retrieval is cheaper but takes 48 hours. Glacier Flexible offers expedited retrieval in minutes at $0.03 per GB.
A team paying around $3 per month to store 1.8TB in cold storage saw the retrieval bill come to roughly $50 the day they needed it back, more than ten times the monthly storage cost. Scale that to 80TB at expedited rates, and a single urgent recovery job runs $2,400 before compute, egress, or engineering time are added to the bill.
The bar is simple: archive storage only pays off for data you genuinely will not need for at least six months. If recovery needs to happen under any kind of pressure, archive tiers cost more than they save. Good means matching storage tier to actual access patterns, not paper pricing.
How to fix the three charges
Each of these charges has a structural cause that ad-hoc scripts cannot solve at scale. Closing them requires a control plane that keeps backup posture aligned with the environment as it grows.
CBPM is that mechanism. We built it because teams could not see what was protected, over-retained, or drifting in their cloud environments.
Eon's approach targets each cost driver separately:
- Storage architecture savings: One hot tier for all backup data eliminates tier management complexity. Deduplication and incremental capture deliver 30-50% smaller backup footprints, directly reducing storage costs.
- Policy enforcement savings: CBPM classifies every resource and applies retention that matches the data type. Low-value data stops sitting on high-cost schedules. The result: right-sized protection across your entire environment.
- Operational efficiency savings: Autonomous discovery picks up new workloads automatically. Granular recovery pulls back individual files, records, or tables instead of full volumes, eliminating the manual work that drains engineering hours.
- Pricing model savings: Agentless deployment means no added compute or infrastructure. No restore fees and no API charges; pricing is tied to the storage footprint, not the activity against it.
Innago implemented these changes, cutting backup costs by 40%. They replaced traditional snapshots with Eon’s backup-optimized storage tier. Restore times dropped to under 15 minutes. SOC 2, PCI, and GDPR compliance stayed intact throughout.
SoFi saw similar results. Before switching to Eon, their team dealt with tickets, manual snapshots, and recovery times that stretched to a full day.
After rolling out Eon across five AWS regions, recovery time dropped from a full day to under five minutes. CJ Keefe, Director of Corporate Infrastructure, DevOps & SRE, at SoFi, reported over 100% ROI in the first year.
Good looks like this: you can answer three questions about any backup at any time: what's protected, how long you're keeping it, and exactly what it costs to restore. If you can't answer all three instantly, the environment isn't under control.
For teams that want to start immediately:
- Audit snapshot sprawl across every account, identifying orphaned snapshots, unattached volumes, and snapshots tied to retired workloads.
- Apply lifecycle policies to every versioned bucket, setting expiry rules that match your actual recovery window.
- Match storage tier to access pattern. If your rotation cycle is shorter than the archive tier's minimum retention window, calculate the real cost and decide whether the savings hold up.
- Standardize policy across clouds. Mismatched retention rules between AWS, Azure, and Google Cloud create both compliance gaps and unpredictable bills.
A leakproof backup environment
When backup posture is automated and visible, three things change immediately:
- The bill becomes predictable because lifecycle policies, correct tier selection, and granular restores eliminate the charges that appear unexpectedly mid-month.
- Audits take hours instead of weeks because classified, policy-driven data with a central view across accounts gives auditors exactly what they need without manual digging.
- New resources are protected from day one, because automated discovery means nothing runs unprotected, even when someone forgets to update a script.
Want to see exactly where your cloud backup costs are going? Book a free demo, and we will walk through how Eon maps your environment and surfaces what is driving spend.
Frequently asked questions
How much does cloud backup cost per month?
Cloud backup costs $2 to $47 per month for 2TB at the storage layer, depending on the provider and tier. At 500TB on AWS, the first-month bill can exceed $25,000 once storage, API calls, and data transfer fees are included.
What are the main hidden costs of cloud backup?
The main hidden costs of cloud backup are minimum retention penalties, full-restore egress fees, and unmanaged object versioning charges. Most teams find all three during a billing audit, well after the charges have already run up.
Why do cloud backup costs keep rising?
Cloud backup costs keep rising because backup environments grow faster than the visibility tools meant to manage them. Snapshots accumulate, object versions stack up, and retention windows extend, all without showing up as separate line items on the bill.
Is AWS Glacier really the cheapest option for long-term backup?
No, AWS Glacier is not always the cheapest option once retrieval costs are factored in. Storage runs $0.00099 per GB per month, but expedited retrieval costs $0.03 per GB. Glacier only pays off for data you genuinely will not need for at least six months.
How can I reduce cloud backup costs without losing coverage?
You can reduce cloud backup costs without losing coverage by auditing snapshots for over-retention, applying lifecycle policies to versioned buckets, and matching storage tier to actual access patterns. Combining all three typically delivers the biggest monthly savings.
What is the difference between cloud backup and cloud storage?
Cloud backup creates protected recovery copies of your data for restoring after data loss, corruption, or ransomware. Cloud storage syncs files for access across devices. Consumer tools like Google Drive and Dropbox are sync tools, not backup solutions.
How does Eon's storage approach differ from native cloud backup?
Eon uses a purpose-built storage tier that deduplicates and compresses backup data, typically reducing storage footprints by 30-50%. Native cloud backups store full snapshots and versions at standard rates. The difference shows up directly in monthly storage costs, separate from any restore or operational savings.
