Skip to main content

EBS: Delete Idle Volumes

Updated over a week ago

Difficulty: Medium

Description

We identified EBS volumes in your account that are attached to running instances but have had no read or write activity. Before deleting, a snapshot should be created to ensure data recoverability.


Selection Criteria

  • The EBS volume is attached to a running instance.

  • The EBS volume has less than 1 IOPS/day (VolumeReadOps + VolumeWriteOps) over the last 7 days.


Expected Saving

100% of the annualized EBS cost for the affected volume.

Example: a gp3 volume of 100 GB costs ~$9.60/year — savings are realized immediately upon deletion.


Operational Impact

⚠️ This recommendation involves permanent data deletion. Extra caution is required.

Downtime: None if the volume is detached before deletion. However, if the volume is a root volume or is used by the OS, detaching it will crash the instance.

Breaking change: YES — EBS volume deletion is irreversible without a prior snapshot.

Root volume risk (CRITICAL): Before any action, it is essential to verify whether the volume is the root volume of its attached instance (typically mounted as /dev/xvda, /dev/sda1, or equivalent). Deleting a root volume while the instance is running will cause immediate instance failure and potential data loss. Root volumes must be explicitly excluded or handled with a separate, manual validation workflow.

Recovery: Possible only if a snapshot was created beforehand. Snapshot restoration incurs additional S3 storage costs.

Recommended action: Always create an EBS snapshot before deletion. Communicate to the customer that deletion is permanent and requires their explicit approval.


References:

Did this answer your question?