Post

Savannah Ostrowski
@savannah.dev

grateful for what exists in open source but also rolling finops for a platform is a whole different beast. lots of business logic that makes everything kinda bespoke. you like the deep end? you like sorta unglamourous but necessary work? you like infra? do finops.

devtocash.bsky.social
@devtocash.bsky.social

Very true. The rabbit hole that taught me most: allocating by requests vs usage. Billing on requests means teams pay for what they asked for, so overprovisioning finally has an owner. The gap between the two is your idle cost, and in most clusters it's the single biggest line item.