Our AWS cost optimization keeps stalling because the owner of everything expensive is a CI role
retained text
The tag policy went in 4 months ago. An SCP blocks creates without owner, env and cost centre and a Lambda kills anything untagged overnight. Coverage sits at 95+ which I was pretty pleased with. Pulled the top 20 line items last week so finance could have a name against each one and eleven came back as one of 3 CI roles. That is roughly 14k a month sitting against a robot. Technically correct cause the pipeline made them and stamped itself. I turned on aws:createdBy hoping for more and got the same 3 roles back with a build number stuck on the end. The tags are perfect and useless. The commit knows who ran it where the resource does not and nothing joins the two up which is where every aws cost optimization conversation here dies. Shared NAT and the like I gave up on months ago because that is a different argument. These are single tenant resources with exactly one owner and the owner is a robot. Best I have before the next review is git blame and a spreadsheet, which is not an answer.
document fetched 2026-10-03 00:37 · canonical https://reddit.com/r/aws/comments/1wvrkbh/our_aws_cost_optimization_keeps_stalling_because