S3, EBS and Snapshot Costs: Where Storage Money Leaks

Your Storage Bill Never Shrinks: 6 Ways to Cut S3, EBS and Snapshot Costs

Published by Rushikesh Dhule, LessBill · November 5, 2026 · Updated November 16, 2026

Compute gets all the attention. You rightsize instances, buy Savings Plans, kill idle GPUs. But open your bill a year later and there is one line that only ever goes up: storage.

Storage is sneaky because nothing breaks when it grows. Nobody gets paged because a bucket doubled. Old disks, forgotten snapshots and logs from a project that ended last spring just sit there, billed every single month. The good news: storage waste is some of the easiest money to get back, and most fixes are one-time settings, not ongoing work.

Here are six places to look, starting with the fastest wins.

1. You are still running gp2 volumes

On AWS, gp3 EBS volumes cost about 20% less per GB than the older gp2 type, and they give you a baseline of 3,000 IOPS no matter how small the disk is. With gp2, performance scaled with size, so teams often bought bigger disks just to get speed. That habit is still baked into a lot of launch templates.

Fix: List every gp2 volume and change it to gp3. The switch happens in place, with no downtime and no detaching. Then update your Terraform modules, AMIs and launch templates so new volumes default to gp3.

2. Unattached disks nobody owns

When you terminate an instance, its EBS volume is not always deleted with it. Those "available" volumes keep billing at full price while doing nothing. They pile up fast after load tests, migrations and autoscaling experiments.

Fix: Filter EBS volumes by state "available." For each one, take a final snapshot if there is any doubt, then delete the volume. Set "Delete on termination" to true in your launch templates so this stops happening.

3. Snapshots that live forever

Daily backups are good. Daily backups kept for three years with no expiry are a quiet leak. Standard EBS snapshots cost roughly $0.05 per GB-month, and a busy account can hold tens of terabytes of them.

Fix: Use Amazon Data Lifecycle Manager or AWS Backup to set retention rules (for example, keep 7 dailies, 4 weeklies, 12 monthlies). For snapshots you must keep for compliance but rarely restore, move them to the EBS Snapshots Archive tier, which AWS says is up to 75% cheaper. Note the catch: archived snapshots have a 90-day minimum and take hours to restore, so only archive things you truly will not need quickly.

4. Everything in S3 sits in Standard

S3 Standard is around $0.023 per GB-month in us-east-1. Most data is hot for a few weeks and then almost never read again. Leaving all of it in Standard means paying top price for data nobody touches.

Fix: For buckets with unpredictable access, turn on S3 Intelligent-Tiering. It moves objects to cheaper tiers automatically when they go cold, with no retrieval fees in its standard tiers. Two things to know: it charges a small monitoring fee per object, and objects under 128 KB are never moved down. So for buckets full of tiny files, it may not help. For data with a clear life cycle (logs, exports, backups), use lifecycle rules to move data to Glacier classes after a set number of days, and expire it when it is no longer needed.

5. Hidden junk inside your buckets

Two things bloat S3 without showing up in a normal file listing. First, incomplete multipart uploads: failed large uploads leave parts behind, and you pay for them. Second, old object versions: with versioning on, every overwrite or delete keeps the previous copy, sometimes forever.

Fix: Add a lifecycle rule to every bucket that aborts incomplete multipart uploads after 7 days. On versioned buckets, add a rule that expires noncurrent versions after 30 to 90 days. Use S3 Storage Lens to find which buckets are growing fastest.

6. Nobody knows whose storage it is

The deeper problem behind all of the above is ownership. If a bucket or volume has no owner tag, nobody feels safe deleting it, so it stays forever "just in case."

Fix: Require owner, team and environment tags on storage resources, and enforce them with tag policies or your IaC checks. Review untagged storage monthly. Anything untagged and untouched for 90 days goes on a delete list.

Worked example: two fixes with the numbers

gp2 to gp3. AWS's EBS pricing page uses $0.10 per GB-month for gp2 and $0.08 for gp3 in its examples. For 2,000 GB of volumes: 2,000 x $0.10 = $200 a month on gp2, 2,000 x $0.08 = $160 on gp3. That is $40 a month, $480 a year, and gp3 includes a 3,000 IOPS and 125 MB/s baseline. Find the candidates and convert one:

aws ec2 describe-volumes --filters Name=volume-type,Values=gp2 --query "Volumes[].[VolumeId,Size]" --output table

aws ec2 modify-volume --volume-id vol-0123456789abcdef0 --volume-type gp3

Check first whether a large gp2 volume was relying on more than 3,000 IOPS; if so, set the IOPS you need on gp3 (extra IOPS are billed separately).

Snapshot archive. 5,000 GB of standard snapshots at $0.05 per GB-month costs $250 a month. Move the 4,000 GB you keep only for compliance to the archive tier at $0.0125: 4,000 x $0.0125 = $50, plus 1,000 x $0.05 = $50 for the recent ones. New total $100, a saving of $150 a month.

When archiving is the wrong move: the archive tier has a 90-day minimum, and an archived snapshot has to be restored before you can use it. Keep anything you might need for a fast recovery in the standard tier. For objects in S3, the same idea is automated with lifecycle rules.

Related guides: 7 Zombie Cloud Resources Quietly Draining Your Bill (and How to Kill Them) and Who Owns This Cloud Bill? A 5-Step Tagging Plan That Makes Every Dollar Traceable.

The bottom line

Storage costs rarely spike. They creep. That makes them easy to ignore and easy to fix. Switch gp2 to gp3, delete orphaned disks, put retention on snapshots, tier your S3 data, clean up multipart uploads and old versions, and tag everything with an owner. Most of these are one afternoon of work and keep paying back every month after.

How LessBill can help

LessBill.org scans your AWS, GCP and Azure accounts for storage waste like unattached disks, old snapshots, gp2 volumes and untiered buckets, shows you exactly what each one costs, and helps your team clean it up safely. Start with a free scan and see how much your storage bill could shrink.

Sources and further reading

Also drawn on: AWS News Blog (EBS Snapshots Archive), Amazon EBS and S3 documentation and pricing pages, CloudForecast, DoiT.

Comments

Popular posts from this blog

Stop Paying On-Demand Prices: Savings Plans, Reserved Instances and Spot Explained

7 Zombie Cloud Resources Quietly Draining Your Bill (and How to Kill Them)

When Does Azure SQL Serverless Beat Provisioned? The Maths