Cross-stack FinOps: bill anatomy, order-of-magnitude war stories, chargeback culture — the capstone remix.
The cross-cutting capstone, told in primary sources: Netflix's Byte Down attribution program, Uber compressing its logging bill by two orders of magnitude, Wix halving platform spend with chargeback, Canva's $300K/month storage-class arithmetic — and the mechanism classes (representation, substrate, pricing model) that separate order-of-magnitude cuts from knob-turning.
Six cuts from 45% to 99%+, classified by mechanism: change the representation (Uber's 169x, the 8,970-to-1.4 slot-hour collapse), the substrate (PayPal's GPUs, Wix's EKS), or the pricing model (Foodpanda's slots, DoorDash's disk-priced store) — never just the knob.
Uber's Spark cluster emits up to 200TB of unstructured logs a day — 5.38PB per 30 days — from 250K+ daily jobs. Storing one month on HDFS priced out at ~$1.8M/year, so the team did what every team does first: they cut retention to three days and dropped verbosity from INFO to WARN. Debugging got harder; the bill was still ~$180K/year; and the small-write pattern was burning through SSDs. That's the shape of a cost problem that knob-turning can't solve — the *workload* is the cost, and no retention slider changes what a log line weighs. This week is a tour of teams who stopped turning knobs: every story cuts 45% to 99%+, and every one does it by changing the representation, the substrate, or the pricing model.
The full week 2 brief is part of LeetData Pro.
