Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Evaluate eBPF and how it can be applied (or not) #956

Open
JorTurFer opened this issue Mar 26, 2024 · 10 comments
Open

Evaluate eBPF and how it can be applied (or not) #956

JorTurFer opened this issue Mar 26, 2024 · 10 comments

Comments

@JorTurFer
Copy link
Member

JorTurFer commented Mar 26, 2024

Proposal

Currently, we are deploying the interceptor instances as containers, which works well, but eBPF could improve the performance working at kernel level instead of at application level.

Using eBPF we should be able to get the metrics about the traffic, so technically we should be able to get all the metrics that we need for scaling but we still have to check if we can "hold" the request during cold starts.

Actually, this issue is just a place to bump thoughts to plan next steps in this direction than a real issue or feature request. We should define the next steps from here.

@isala404
Copy link

Hi,
We are using the KEDA HTTP add-on to provide scale-to-zero functionality for our internal platform, and one day, this exact thought popped into my mind. So, I just made a really quick POC using Rust and Aya, and the results were promising.

https://github.com/MrSupiri/scale-to-zero-ebpf

Copy link

stale bot commented May 27, 2024

This issue has been automatically marked as stale because it has not had recent activity. It will be closed in 7 days if no further activity occurs. Thank you for your contributions.

@stale stale bot added the stale All issues that are marked as stale due to inactivity label May 27, 2024
@zroubalik zroubalik removed the stale All issues that are marked as stale due to inactivity label May 28, 2024
Copy link

stale bot commented Jul 27, 2024

This issue has been automatically marked as stale because it has not had recent activity. It will be closed in 7 days if no further activity occurs. Thank you for your contributions.

@stale stale bot added the stale All issues that are marked as stale due to inactivity label Jul 27, 2024
Copy link

stale bot commented Aug 4, 2024

This issue has been automatically closed due to inactivity.

@stale stale bot closed this as completed Aug 4, 2024
@github-project-automation github-project-automation bot moved this from To Triage to Done in Roadmap - KEDA HTTP Add-On Aug 4, 2024
@mmalyska
Copy link

It's sad that it got closed without any response. I'm really interested in this with usage of tcp/udp traffic to scale my services to 0 when no customer traffic is detected.

@mmalyska
Copy link

@JorTurFer can we reopen this?
@isala404 did you had any success in using this with cilium and keda?

@JorTurFer JorTurFer reopened this Dec 15, 2024
@stale stale bot removed the stale All issues that are marked as stale due to inactivity label Dec 15, 2024
@JorTurFer
Copy link
Member Author

It's sad that it got closed without any response. I'm really interested in this with usage of tcp/udp traffic to scale my services to 0 when no customer traffic is detected.

That's currently possible with the current approach without eBPF

@JorTurFer can we reopen this?

Sure!

@mmalyska
Copy link

It's sad that it got closed without any response. I'm really interested in this with usage of tcp/udp traffic to scale my services to 0 when no customer traffic is detected.

That's currently possible with the current approach without eBPF

@JorTurFer can we reopen this?

Sure!

I cannot find anywhere how to scale TCP or UDP workloads with current approach. I can only see HTTP ones using ingress.

@isala404
Copy link

@isala404 did you had any success in using this with cilium and keda?

Sorry, I couldn't find time to improve upon my POC. I will try to share some more results in a few weeks.

@JorTurFer
Copy link
Member Author

I cannot find anywhere how to scale TCP or UDP workloads with current approach. I can only see HTTP ones using ingress.

True, I misunderstood the requirement :/ sorry

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
Status: Done
Development

No branches or pull requests

4 participants