gp2 vs gp3 EBS: How to Cut EBS Storage Costs ~20%

If you still run gp2 EBS volumes, migrating to gp3 is one of the lowest-risk, highest-return changes available on AWS. Here's what changed and how to do it safely.

What's the difference?

gp2 ties performance to volume size: you get 3 IOPS per GB, so the only way to get more performance is to provision more storage than you need. gp3 decouples them — you pay for capacity, and IOPS and throughput are configured separately, with a generous free baseline (3,000 IOPS and 125 MB/s) included at no extra cost.

On price, gp3 storage is roughly 20% cheaper per GB than gp2. For the large majority of general-purpose volumes, gp3 is cheaper and at least as fast.

When gp3 wins

• General-purpose workloads within the 3,000 IOPS / 125 MB/s baseline — pure savings.
• Volumes you over-sized purely to get more gp2 IOPS — you can shrink the IOPS-for-size tax.
• Workloads needing high throughput at moderate size — gp3 lets you buy throughput directly.

When to look closer

If a volume needs sustained IOPS well above the baseline, price out the extra provisioned IOPS and throughput on gp3 versus gp2 — gp3 is still usually cheaper, but confirm for very IOPS-heavy databases. For latency-critical, very high-IOPS workloads, io2 may be the right tier regardless.

How to migrate (no downtime)

1. Snapshot the volume first as a safety net.
2. Change the volume type to gp3 in place — EBS Elastic Volumes lets you modify type, IOPS, and throughput on a live, attached volume.
3. The volume enters an optimizing state but stays fully usable throughout.
4. Leave IOPS/throughput at the free baseline unless monitoring shows you need more.

There's no need to detach, no reboot, and no data movement on your side. The change is reversible if you ever need to go back.

Doing it at scale

One volume is easy; finding every gp2 volume across accounts and regions is the hard part. CloudWireHub flags all of your gp2 volumes and estimates the monthly savings of moving each to gp3 — try it free.