Hacker News new | ask | show | jobs
by turtleyacht 1016 days ago
Why don't we have a tool that lets us associate monetary budgets to concrete hardware specs, and then generates a cross-cloud (or cloud-agnostic) configuration?

  budget: $23000
  pricing: cheapest
  cpu:
    preferred: xeon
    cores: 8
    mem: 64gb
  disk:
    preferred: ssd
And some --import-usage flag to specify a folder of historical CPU and memory usage from Grafana.
2 comments

This is backwards. What you want is to specify resource limits and then have the maximum possible bill calculated for you.

If you want services to turn off if they exceed a threshold then you will need some sort of priority scheme.

A capped resource limit and/or budget does sound easier to manage. Here is your possible max; after that, shut stuff down in this order.

If it were any easier, it would likely be automated as a layer above Infra as Code (IaC). Someone may need to manually check on teams and ensure critical apps are not throttled before less critical ones.

Found a great option for you at $23000. Amazingly we managed to accommodate your budget. /s
It does sound foolish to share that directly to the cloud provider. They have an incentive to maximize spend--although they may also provide more than enough oomph with that much spend. It would lie on the payor to collect on all the available discounts (multi-year commitment, etc).

Assuming the tool is more or less one step removed from an elaborate spreadsheet and more of a crowdsourced, open optimizer, do you feel there would be any gains to generating multi-cloud setups, or just unnecessary complexity?

Is it possible to have a good-faith network of operators reporting monetary info along optimal routes?

Routes less like distances, and more like "given your compute and memory profile, a tier 2 storage bucket in AWS and 4gb VM in GCP would fit your budget."

This is a nice reminder of the importance of having your software be a "user-agent"!
Obviously you don't want to tell AWS what your budget is. You want your local tool that does work for your benefit to have a way to sanity check the configuration, and warn you if the only suitable VM at the provider turns out to cost a lot more than you expected.
Thank-you.

> local tool... to sanity check

Yes, exactly. Some open-source project as a layer between the cloud providers, and budget and other elements are just knobs to twiddle.

If this "open pricing" project could swarm various known constraints between money and virtual hardware, would it be a kind of constraint satisfaction problem?