Skip to main content
emnode
Sign up
Learning path

Kill idle waste

Unattached volumes, idle endpoints, orphaned IPs and empty tables: delete what nothing uses.

9 lessons·~112 min total

Lessons in this path

  1. 1
    Cost AWS

    Delete idle VPC endpoints

    An interface VPC endpoint bills per hour per AZ even with zero traffic: a 3-AZ endpoint is ~$22/month for nothing. Find the idle ones and delete them (but never the free gateway kind).

    13 min
  2. 2
    Cost AWS

    Delete unused DynamoDB tables

    A DynamoDB table with no reads keeps billing for storage, reserved capacity, backups, and replicas, find the dead ones, archive their data, and delete them before the layers compound.

    13 min
  3. 3
    Cost AWS

    Delete unused EFS file systems

    An idle EFS file system has zero IO and zero connections, but it bills every gigabyte stored, every month, at a rate that dwarfs S3 and EBS.

    13 min
  4. 4
    Cost AWS

    Delete unused Elastic Network Interfaces

    A detached ENI isn't billed on its own, but it can hold a public IPv4 you're paying for, and it blocks the cleanup of everything attached to it. Find the true orphans and delete them safely.

    12 min
  5. 5
    Cost AWS

    Release unattached Elastic IPs

    An Elastic IP that isn't attached to anything still bills ~$3.65/month, and since 2024 even the ones in use cost money. Find the orphans and release them safely.

    11 min
  6. 6
    Cost AWS

    Delete idle ElastiCache clusters

    An ElastiCache cluster serving no traffic still bills per node-hour for its full reserved capacity: find the idle ones, take a final snapshot, and tear them down cleanly.

    13 min
  7. 7
    Cost AWS

    Delete empty CloudWatch log groups

    An empty log group costs almost nothing directly, but hundreds of them obscure what's actually logging, and most carry no retention waiting to bill silently.

    11 min
  8. 8
    Cost AWS

    Scale in over-provisioned Auto Scaling groups

    An ASG with a min capacity set high 'for safety' runs idle instances 24/7, tie capacity to a real demand signal instead of a static guess.

    14 min
  9. 9
    Cost AWS

    Reduce SQS queue message retention

    Queues default to a long retention "just in case"; tighten it once you know the real consumer SLA.

    12 min