Add example for kubeflow spot instance cluster config

[arjun921]

Helps users build a highly cost efficient Kubeflow cluster for ML/DL training at 50-80% less cost

Description

Added cost efficient Kubeflow cluster spec with

• Spot GPU instances
• Scale down to zero

Checklist

• Added tests that cover your change (if possible)
• Added/modified documentation as required (such as the README.md, or the userdocs directory)
• Manually tested
• Added labels for change area (e.g. area/nodegroup), target version (e.g. version/0.12.0) and kind (e.g. kind/improvement)
• Make sure the title of the PR is a good description that can go into the release notes

[martina-if]

Hi @arjun921 , very nice and complete example! Thank you!
I only have one nitpick and it's that it would be nice to split long lines (line number 17, 93 etc) into shorter ones.

Other than that I think this can be rebased, approved and merged :)

[arjun921]

Hi @martina-if ,

That's a really good catch!

I'll make those changes and push soon.

I have one doubt though, would it be helpful if I included a detailed blog post as a comment in the file?

Here's the blog:
https://blog.gofynd.com/how-we-reduced-our-ml-training-costs-by-78-a33805cb00cf

I was thinking if it helps, I could include it like line 3 on the source:
https://github.com/arjun921/aws-spot-instances-kubeflow/blob/master/envs/staging/cluster-spec.yml

[martina-if]

Yeah, I think the link could be useful for users 👍

[arjun921]

@martina-if All requested changes done!
Please let me know if there's anything that I might have missed out on.

[arjun921]

@martina-if I've updated test cases,
it's hopefully ready for a merge 😬