Beta

Reports What to change

AWS waste

Money paid for what is not used, sorted by how easily it goes: unused, overprovisioned or stranded, by category and by service. One figure for what can be saved, and honest figures for what cannot.

What it shows

A sentence at the top: how much a month can be saved, how much more is idle but held in place, how much is kept as headroom on purpose and how much could not be judged. Four figures repeat it. Under Lines, every entry by category: unused, which nothing depends on and can be deleted; overprovisioned, in use but with something smaller or fewer that fits today; stranded, idle but held in place by a named blocker, such as a reservation or a dependency; headroom, kept on purpose; and not measured, cost nothing could judge. Each line names the resource, its kind, account and environment, the blocker if any, what to do and why. A trend by category and bars by service show how the waste moves.

The Kubernetes lines are the idle capacity of each cluster, from the Kubernetes cost report's split of each node: the EKS idle capacity you pay EC2 for, or the idle part of a cluster on your own servers at the price you set. A cluster of your own that sent nothing in the 7 days is listed as not measured, with no amount. ECS clusters on EC2 are measured the same way: the spare capacity of their ECS instances, and what each ECS service reserves and leaves unused. A service's use is read from CloudWatch only with the usage metrics option on; without it, its tasks are listed as not measured.

Questions it answers

  • How much of our AWS bill is waste, and how much of that can actually be removed?
  • What is idle but stuck, and what is holding it?
  • How much EKS idle capacity are we paying for across clusters?
  • Is the waste shrinking month over month?

What it reads

Nothing new is read from AWS for this page. Once a day, after the savings recommendations, what the other reports found is sorted by how removable it is. Money is per month at list price over the last 7 days, or at the price set for a cluster on your own servers, so the bill will show less and a report younger than 7 days shows less than a full month. Spikes shorter than an hour do not show.

What to do with it

Read the "can be saved" figure as the target and the unused category as the first week's work. Stranded money is never counted as a saving: it tells you what a blocker costs, which is the number to bring to the conversation about removing it. Headroom is a decision already made, listed so that it stays a decision. The changes themselves, with commands where one does it, are on the savings recommendations page; the Kubernetes idle cost by cluster and namespace is in Kubernetes cost.

Related reports

  • AWS savings recommendations: Changes that would cut the bill, priced per month, largest first, and what the changes made so far save.
  • Kubernetes cost: What each namespace and workload costs on your Kubernetes clusters, in AWS and on your own servers, and a month at this rate, by namespace and by cluster.
  • AWS infrastructure findings: Unattached volumes, idle Elastic IPs and versions past support, by rule and by account, with the resources that fail.
  • AWS resources: Every resource the inventory found, with its cost, by type and by its Environment and Team tags.

Every plan has every report, hosted or self-hosted, and the hosted plan is free during early access: see pricing. Like every report, this one splits and filters by account, environment, service and resource, steps back 13 months, downloads as PDF and CSV and keeps the view in its link: what every report does. All reports.