terraform-provider-iterative VS carbonifer

You should look at terraform, and at the provider from iterative, the guys begind DVC, https://registry.terraform.io/providers/iterative/iterative/latest/docs

TPI tool offers full lifecycle management of computing resources (including GPUs and respawning spot instances) from several cloud vendors (AWS, Azure, GCP, K8s) without needing to be a cloud expert: Terraform Provider Iterative (TPI) - Terraform Registry

Terraform Provider Iterative (TPI) address the specific needs of machine learning teams - it is an open-source tool extending the functionality of Terraform, the world's most widely used multi-cloud provisioning product. The tool enables full lifecycle management of computing resources and is designed specifically for machine learning pipelines: Terraform plugin for machine learning workloads: spot instance recovery & auto-termination | AWS, GCP, Azure, Kubernetes

TPI is a Terraform plugin built with machine learning in mind. This CLI tool offers full lifecycle management of computing resources (including GPUs and respawning spot instances) from several cloud vendors (AWS, Azure, GCP, K8s) without needing to be a cloud expert: Terraform Provider Iterative (TPI) - GitHub

Hey all, we (at https://iterative.ai) just launched TPI - Terraform Provider Iterative: https://github.com/iterative/terraform-provider-iterative It was designed for machine learning (ML/AI) teams and optimizes CPU/GPU expenses.

I've started to work on Carbonifer, a tool that can estimate carbon emissions before deploying an infrastructure. A bit appreciate infracost, this reads Terraform files and estimates carbon emissions if this plan is applied. Currently targeting GCP computes, I plan to expand it to other providers and resources and of course target other IaC (pulumi? Chef?...). This is currently using average carbon intensity data (gCO2eq/kWh) but this will also read real-time data (example of real-time‏‏‎‏‏‎‏‏‎‏‏‎­data source: electricitymaps). Carbonifer can help to pick the best machine type/region/provider but I also can imagine it coupled with a scheduler to deploy where/when it's less carbonated or even resize/move an existing infra. Have a look at https://github.com/carboniferio/carbonifer I would appreciate any feedback or thoughts about it. Do you have this kind of need?

Some tools like Carbonifer reads terraform files and estimate carbon emissions, which varies a lot depending on your provider region. That tool could be part of your CI to show what would be the carbon impact of a change request.

Have a look at https://github.com/carboniferio/carbonifer

Reads Terraform files and estimate what will be the carbon emissions of a infra project if deployed. Currently supports only GCP (AWS, Azure are work in progress).

In the future, it will suggests better machine types or providers regions to minimise carbon footprint and will help pilot carbon footprint of infra with realtime carbon emissions of the grid.

Github: https://github.com/carboniferio/carbonifer