[punkrex]

A lot of the “finops” practitioners I’ve seen are myopically focused on tagging of AWS resources; and that falls to pieces with kubernetes because AWS can’t see inside the kubernetes clusters.

I’m not surprised they don’t like it.

[sleepybrett]

I just had a meeting with a 'finops' manager where i showed him how kubernetes has a similar tagging structure (labels) and how we can break down per team pricing based on cpu/memory utilization.

It's not hard, you just need the tools (kubecost, etc)

[cobookman]

pod A uses 2 cores, pod B uses 1 core. Machine has 4 cores and all remaining unscheduled pods require 2 cores.

How do you attribute the partial usage of the node? Is it 2 cores billed to pod A, 1 core billed to pod B, and 1 core billed to some random team?

Or do you have 2/3 of Node billed to pod A and 1/3 of Node billed to Pod B.

Now deal with this permutation across all the various variables.

[orf]

You do it roughly based on the deployment requests and the average HPA values throughout a time slice.

Most k8s workloads run on a homogenous set of node types, so you can have an hourly cost per gb and per vcpu without digging too much into it.

[cogman10]

Yup. Particularly because k8s scales based on whole system load, not single app load. It's harder to predict because it's better at optimizing resource utilization and ultimately lowering costs.

[briffle]

You can scale load on many things in more recent versions of K8s. for example, pubsub depth of un-acked messages, or custom metrics in prometheus format.

https://cloud.google.com/kubernetes-engine/docs/tutorials/au...