Why it matters
gp2 is still the default in many templates and older accounts, so it is often one of the largest and lowest-risk savings on an AWS bill. The change happens online, with no downtime.How nOps finds it
nOps reads your EBS billing for the last 30 days. A gp2 volume is flagged when it was billed on at least 28 of those days and gp3 would save at least $1 a month after provisioning the same baseline IOPS (3 IOPS per GB, up to 16,000) and throughput. An io1 or io2 volume is flagged when it has 16,000 provisioned IOPS or fewer and gp3 with the same IOPS would save at least $5 a month.Before you start
Permission to modify EBS volumes (ec2:ModifyVolume) in the volume’s account.
How to implement
- Check the volume’s current type, size and performance, and that no other modification is in progress.
- Change the volume type to gp3, setting IOPS and throughput to at least the values shown in the recommendation (AWS console: EC2 → Volumes → Modify volume, or
aws ec2 modify-volume --volume-type gp3). - Wait for the modification to reach optimizing or completed. The volume stays attached and in use throughout.
- If the volume is managed by Terraform or CloudFormation, update the template too, or the next deployment can change it back.
Risks and considerations
- An EBS volume can be modified up to four times in a rolling 24 hours, and each change must finish before the next one starts.
- For io1 and io2, confirm the workload does not depend on io2’s higher durability (99.999%) or sub-millisecond latency before changing.
- To roll back, modify the volume back to its original type and performance.
FAQs
Will the volume be slower after the change?
Will the volume be slower after the change?
No. nOps sizes the gp3 IOPS and throughput to at least what the volume has today, and the estimate includes the cost of that performance.
Do I need to stop the instance?
Do I need to stop the instance?
No. The change is made online, with no detach or restart.