Six ways to cut EBS backup costs without touching cold tier

Here is a scenario that plays out more often than AWS would like to admit. An engineer notices the per-GB price on AWS Backup cold storage. $0.0125/GB-month versus $0.05/GB-month for warm. A 75% discount. They run the math on a napkin, flip the switch on cold transition in their backup plan, and close the laptop feeling like they just found free money.

Six months later, the backup line item on the AWS bill has gone up.

In Part 1 of this series, I walked through exactly why this happens and derived the break-even change rate (18.2% monthly block change, a threshold most production volumes never cross). The short version: cold tier converts every incremental EBS snapshot to a full snapshot, so the 75% per-GB discount gets overwhelmed by the volume explosion. EFS and DynamoDB genuinely benefit. EBS almost never does.

This post is about what to do instead. Six alternatives, each sidestepping the full-snapshot conversion problem, with exact pricing, worked examples, and honest limitations.

1. AWS Data Lifecycle Manager: the free fix hiding in plain sight

This was the first alternative I tried after discovering the cold tier trap, and for most teams it's the only one you need. If your backup scope is EBS volumes and EC2 AMIs, AWS Data Lifecycle Manager (DLM) is almost certainly your answer.

The service is free. No per-policy charge, no management fee. You pay only the underlying EBS snapshot storage rate of $0.05/GB-month in us-east-1. That is the same rate you pay today for warm-tier AWS Backup snapshots.

The difference that matters: DLM manages snapshots directly in the EBS standard tier. Snapshots stay incremental throughout their entire lifecycle. There is no vault abstraction, no tiering mechanism, nothing to trigger a conversion to full snapshots. Cross-region copies created by DLM are also incremental. Cross-account sharing works through event-based policies.

The math makes the case cleanly. Cold tier only wins when your incremental snapshot size exceeds 25% of the full volume. That is because $0.05 × 0.25 = $0.0125, which is exactly the cold tier per-GB rate. Below that threshold, keeping snapshots incremental in warm storage beats cold every time.

For a 1 TB volume with 5% monthly change:

That is a 5x cost difference per snapshot. Multiply that across 50 servers with 2 TB volumes each and 36 monthly snapshots retained, and you are looking at roughly $13,750/month with DLM versus $45,000/month with cold tier. A $375,000 annual difference. And the DLM approach is also faster to restore from.

DLM's limitations are narrow but real. It covers only EBS volumes and EC2 AMIs, with no support for RDS, EFS, DynamoDB, or cross-service backup centralization. It does not support Vault Lock (WORM compliance), restore testing automation, or AWS Organizations-level policy management. Snapshot timing can drift up to 59 minutes from schedule, which matters for tight RPO requirements. Teams with multi-service backup or compliance audit trail requirements still need AWS Backup, just not the cold tier.

2. Reduce backup frequency in warm tier: the fix nobody bothers to try

I almost didn't include this one because it feels too obvious. But when I ran the numbers, it turned out to be one of the biggest levers available.

The math consistently shows that fewer warm-tier snapshots outperform more frequent cold-tier snapshots, and not by a small margin. The full-snapshot conversion is so costly that the expensive tier with fewer copies beats the cheap tier with more copies, almost every time.

Take a 1 TB volume with 500 GB used data and a 20% monthly block change rate:

Strategy C is the one most teams enable when they think they are optimizing costs. It is also the most expensive of the four. It costs 14% more than just keeping everything warm, and 78% more than quarterly warm with no cold tier at all.

Strategy B (quarterly warm) cuts costs by 36% with zero architectural complexity. Instant restores. No retrieval fees. No minimum retention lock. No conversion penalty. You are not giving anything up. You are just backing up less frequently, which is fine if your RPO supports it.

You might notice Strategy D (quarterly + cold) is actually the cheapest at $525. That's because quarterly backups produce fewer snapshots to convert to full, so the cold tier penalty is smaller. But it still carries the 90-day lock-in and 4+ hour restore delay. For most teams, Strategy B gives you 85% of the savings with none of the operational constraints.

The objection most engineers raise here is compliance. "We need monthly backups. Our SOC 2 audit requires it." This turns out to be a myth more often than not. None of the four major frameworks mandate a specific backup frequency:

SOC 2 requires a documented RPO/RTO and evidence you meet it. A weekly or monthly cadence with proper testing passes. ISO 27001 Annex A 8.13 says backups should follow "the agreed topic-specific policy based on risk assessment." PCI DSS v4.0 mandates minimising cardholder data retention. Its 1-year/90-day-accessible rule applies to audit logs, not data backups. HIPAA requires retrievable exact copies of ePHI, specifies no frequency, and its 6-year mandate is for compliance documentation, not backup copies.

The real compliance requirement is that you document your backup policy, test it, and meet your stated RPO. Most teams over-provision because a previous engineer defaulted to "daily, 1 year" and nobody questioned it. Right-sizing frequency to actual RPO requirements and data classification can cut backup costs 40–80% before you touch a single pricing tier.

3. Cross-account snapshot copies: account isolation without the cold-tier penalty

A lot of teams turn to cold storage not for the cost savings but for the account isolation. Separate account, ransomware protection, backup data that is structurally harder to reach from a compromised primary account. Cold storage just happens to be how AWS Backup surfaces that pattern.

When I talk to teams about this, I always ask: are you using cold tier for the cost savings or for the isolation? If it's isolation, there is a way to get it without the conversion penalty.

Copying EBS snapshots to a separate archive AWS account within the same region maintains the incremental chain and avoids the full-snapshot conversion entirely. Same-region cross-account copies carry no data transfer charge. You pay only the standard $0.05/GB-month on the incremental data in the destination account.

For the same 500 GB volume with 25 GB incrementals:

That is an 80% saving, with faster restores and no 90-day lock-in.

AWS Backup natively supports this through cross-account copy rules in backup plans. You keep the centralized management, Vault Lock, and compliance audit features. You just target a warm vault in an archive account instead of transitioning to cold.

There are two hard requirements. First, customer-managed KMS keys are mandatory. Cross-account snapshot sharing does not work with the default aws/ebs key. KMS key policies must explicitly grant cross-account decrypt permissions to the destination account. Second, incrementality is preserved only when the previous copy in the chain exists and has not been archived. Deleting or archiving the most recent copy in the destination triggers a full copy on the next run. Snapshot lineage management needs automation. Do not rely on manual cleanup.

4. S3 Intelligent-Tiering and S3 Glacier Direct: 68–92% cheaper than cold tier

This is where things get interesting for teams with serious retention requirements. AWS Backup cold storage charges $0.0125/GB-month and calls it cheap. S3 Glacier Deep Archive charges $0.00099/GB-month. That is ninety-two percent cheaper. For teams retaining terabytes for years, that gap is not marginal.

The catch is that EBS snapshots do not natively live in S3. Getting data there requires exporting AMIs as binary objects using AWS's CreateStoreImageTask API. The export creates full images, not incrementals, so you give up the incremental storage efficiency. Whether the deal still works depends on your retention horizon.

For a 500 GB volume held for three or more years:

For long-dated compliance archives, the "keep one verified snapshot of this database at year-end for seven years" use case, Glacier Deep Archive wins by a wide margin even after accounting for the full-image export.

S3 Intelligent-Tiering is a lower-friction middle option. Objects that sit untouched automatically land in the Archive Instant Access tier at $0.004/GB-month (68% cheaper than cold tier) with millisecond retrieval and no retrieval fees. There is no minimum storage duration, so the 90-day lock-in disappears. The trade-off is a monitoring fee of $0.0025 per 1,000 objects per month, which adds meaningfully at high object counts. Objects under 128 KB stay in the standard Frequent Access tier at $0.023/GB.

The implementation work is real. If you need this level of archival cost reduction and you have the engineering bandwidth to build the export pipeline, the savings justify it at scale. If you are under 50 TB and under 3-year retention, one of the first three alternatives is probably simpler.

5. Third-party tools: incremental archiving without the AWS conversion penalty

Three tools in the market have built around the exact problem this post describes: archiving EBS data incrementally to S3 or Glacier without the full-snapshot conversion step.

N2WS reads changed blocks directly from EBS using the EBS Direct APIs, then writes only those changed blocks to S3 or Glacier. The original EBS snapshot is deleted after archival via a feature called ZeroEBS, so long-term storage costs drop to S3/Glacier rates on incremental data only. For a 1 TB volume with 3% monthly change retained for a year, ZeroEBS with Glacier can cut storage costs by roughly 92% versus EBS warm tier. Standard pricing starts at $249/month for up to 20 instances. ZeroEBS and Glacier archiving require the Enterprise tier.

Veeam Backup for AWS stores backups in Veeam's own format directly in S3, using a forever-forward incremental approach driven by EBS Changed Block Tracking via the EBS Direct APIs. Data lives in S3 or Glacier as Veeam-format backups and are never treated as EBS snapshots or subject to the archive conversion. The Free Edition covers up to 10 instances. Paid licenses run $250–$450 per workload per year plus AWS infrastructure costs (S3 storage, worker EC2 instances for recovery). The risk is chain dependency: a corrupted full backup compromises all subsequent incrementals.

[Druva](https://www.druva.com) operates as a pure SaaS platform with global source-side deduplication, claiming 20–50x data reduction across volumes with overlapping data. Backup data lives in Druva-managed infrastructure, completely air-gapped from your AWS environment. This sidesteps the EBS snapshot archival problem entirely. I worked at Druva before starting CloudYali, so I know firsthand how their dedup engine works. The limitation: Druva's AWS-native mode (CloudRanger) still uses EBS Snapshot Archive under the hood, so the full dedup benefit only materializes in the Enterprise SaaS platform with custom per-TB pricing. Data sovereignty and opaque pricing are genuine considerations.

Third-party tools make the strongest case at 50+ instances with multi-year retention requirements. Below that scale, the license cost eats into the savings.

6. Manual Glacier archiving: surgical precision over autopilot

This is the approach I'd use for our own compliance snapshots if we needed seven-year retention on a handful of critical databases. Most teams overlook it because it sounds like extra work.

When AWS Backup cold tier transitions a backup chain, it converts every incremental snapshot to a full snapshot and moves them all. You have no say in which ones. The process is automated and expensive.

The alternative is to identify one specific snapshot, typically a clean full snapshot taken at a known-good state like a quarterly or annual checkpoint, and manually archive just that snapshot to the EBS Snapshots Archive tier via the console or CLI. One full snapshot. One cold archive charge. No incremental chain explosion.

For compliance use cases like "keep one verifiable point-in-time snapshot of this production database at year-end for seven years," manual archiving costs $0.0125 per GB per month on a single full copy. A 500 GB volume held for seven years costs $525 total. The same retention policy enforced through AWS Backup cold tier with monthly backups would cost $7,350 for the same period.

The limitation is scalability. Manual archiving across hundreds of volumes requires scripting, and it does not integrate with AWS Backup's compliance reporting or vault audit trails. It works well as a deliberate, targeted archival strategy for a handful of production systems with long compliance retention requirements. It does not replace a backup policy.

Choosing the right path

Six alternatives sounds overwhelming. In practice, the decision is simple.

Start with the simplest fix. Audit your actual backup frequency and retention requirements against what your compliance framework mandates. Most engineering teams have inherited a policy that says "daily backups, keep for a year," written by someone who was being cautious and never revisited. If your actual RPO is 4 hours, weekly backups satisfy it. If your actual compliance requirement is 90 days, keep 90 days, not 365.

If you are EBS-only and do not need cross-service compliance reporting, switch from AWS Backup to DLM. You pay the same per-GB rate. You stop paying the cold-tier conversion penalty. The migration is a few hours of work.

If you need AWS Backup's centralized management and compliance features but want account isolation, implement cross-account copy rules targeting a warm vault in an archive account. You keep the management layer. You lose the conversion penalty.

Reserve S3 Glacier and third-party tools for large-scale, long-retention scenarios where the per-GB savings compound meaningfully over time. The engineering investment makes sense at 50+ TB retained for 3+ years. Below that, simpler fixes deliver equivalent or better savings.

The universal rule across all six alternatives: never transition EBS snapshots to cold storage without first checking your incremental-to-full size ratio. If your monthly changed data is less than 25% of the total volume, cold tier costs more than warm. For most production databases, application servers, and file systems it sits well below that threshold. AWS provides an EBS Snapshot Tier Evaluator that runs this check in minutes.

What I'd check first

Writing this two-part series taught me something I keep coming back to: the most expensive cloud decisions are the ones that look cheap on the pricing page. Cold tier is just one example. The pattern repeats everywhere in cloud billing. A lower unit price that comes with a hidden volume multiplier, a minimum commitment, or an architectural constraint that quietly changes the math.

The challenge with backup costs specifically is that the mistake is invisible until you look for it. AWS Cost Explorer does not flag backup plans that are paying more in cold tier than they would in warm. It does not tell you that your 12-month retention policy could be 6 months and still satisfy your compliance framework. It does not know that your EBS volumes have a 4% monthly change rate and will never benefit from cold storage.

This is exactly the kind of thing I built CloudYali to catch. For AWS customers, CloudYali analyzes your active backup plans, computes the incremental-to-full ratio for each EBS volume, and flags plans where cold tier is actively costing more than warm. It correlates backup frequency against your retained data patterns to surface over-provisioned retention policies. And it models the cost impact of switching to DLM, cross-account copies, or reduced frequency before you make the change. You can explore this further in our AWS cost optimization documentation.

If backup costs feel opaque or you suspect your team is overpaying on cold tier, a 15-minute demo is a reasonable place to start.

This is Part 2 of a two-part series on AWS Backup cold storage. [Part 1](https://www.cloudyali.io/blogs/aws-backup-cold-tier-cheaper-storage-costs-more) covers the mechanics of the cold-tier cost paradox, the exact break-even change rate calculation, and an interactive calculator you can use to test your own numbers.