Re: [Bpf] [Cats] Fw: New Version Notification for draft-elkins-v6ops-bpf-pdm-ebpf-00.txt

Thanks for sharing this with CATS, Nalini.

If you’re looking for an IETF git repo for this, I think it would sit best on one attached to the BPF working group.

Cheers,

Adrian

From: Cats <cats-bounces@ietf.org> On Behalf Of nalini.elkins@insidethestack.com
Sent: 27 February 2024 12:29
To: Cats <cats@ietf.org>; Shihang(Vincent) <shihang9=40huawei.com@dmarc.ietf.org>
Cc: Balajinaidu V. <balajinaidu.191ec111@nitk.edu.in>; Mohit P. Tahiliani <tahiliani@nitk.edu.in>; CHINMAYA SHARMA . <chinmayasharma.201ec113@nitk.edu.in>; AMOGH UMESH <amoghumesh.201cs207@nitk.edu.in>; V6ops List <v6ops@ietf.org>; Dhruv Dhody <dhruv.ietf@gmail.com>; Michael Ackermann <mackermann@bcbsm.com>; bpf@ietf.org
Subject: Re: [Cats] Fw: New Version Notification for draft-elkins-v6ops-bpf-pdm-ebpf-00.txt

Vincent,

> Thanks for sharing. It looks really promising.

Thanks for the kind words.

> Some quick questions:

> 1.       Is the source code open sourced so that we can play it?

We are looking into how to do this.  We can either set up a GitHub repo ourselves at NITK or we can work with v6ops / CATS / BPF.   

Thoughts?

> 2.       Does eBPF requires root permission to load or run?

The actual process of loading the programs into the kernel requires root permissions.  

> I.e, Does any application developer can load/run a eBPF code by itself without the intervention of the OS?

I’m not sure I understood this correctly, but eBPF code, once compiled, will work on all compatible kernels without any problems but it needs root permissions to attach.

BPF folks, please add thoughts, if you have any. 

We are also looking into doing a Windows implementation.

> 3.   It seems the PDM option is a DoH option. Will eBPF works with HBH option? I assume it can handle both.

I believe that any EH (or many other packet modifications) should work without problems.  Again, BPFers, if you know something that we do not, please chime in.

BPF is a great tool!

I don't know if you saw our performance benchmarks comparing our EH implementation in the Linux kernel to eBPF.  It takes slightly longer with eBPF but actually, IMHO, minimally.  And, just to not have to recompile the kernel and all the headaches that causes, we still prefer eBPF.

Thanks,

Nalini Elkins
CEO and Founder
Inside Products, Inc.
https://www.insidethestack.com

President
Industry Network Technology Council
https://www.industrynetcouncil.org

On Tuesday, February 27, 2024 at 04:01:42 AM PST, Shihang(Vincent) <shihang9=40huawei.com@dmarc.ietf.org> wrote: 

Hi Nalini,

Thanks for sharing. It looks really promising. Some quick questions:

1.       Is the source code open sourced so that we can play it?

2.       Does eBPF requires root permission to load or run? I.e, Does any application developer can load/run a eBPF code by itself without the intervention of the OS?

3.       It seems the PDM option is a DoH option. Will eBPF works with HBH option? I assume it can handle both.

Thanks,

Hang

From: Cats <cats-bounces@ietf.org> On Behalf Of nalini.elkins@insidethestack.comSent: Tuesday, February 27, 2024 7:41 PMTo: Cats <cats@ietf.org>Cc: Balajinaidu V. <balajinaidu.191ec111@nitk.edu.in>; Mohit P. Tahiliani <tahiliani@nitk.edu.in>; CHINMAYA SHARMA . <chinmayasharma.201ec113@nitk.edu.in>; AMOGH UMESH <amoghumesh.201cs207@nitk.edu.in>Subject: [Cats] Fw: New Version Notification for draft-elkins-v6ops-bpf-pdm-ebpf-00.txt

Hello List,

You may be interested in a draft we have done describing how to add IPv6 EH using eBPF and comparing the performance of eBPF implementation to EH implementation in the Linux kernel.   This may be useful for CATS.  The link to the draft is below. 

URL:      https://www.ietf.org/archive/id/draft-elkins-v6ops-bpf-pdm-ebpf-00.txt

Thanks,Nalini ElkinsCEO and FounderInside Products, Inc.

https://www.insidethestack.com

President

Industry Network Technology Council

https://www.industrynetcouncil.org

----- Forwarded Message -----

From: Michael McBride <michael.mcbride@futurewei.com>

To: nalini.elkins@insidethestack.com <nalini.elkins@insidethestack.com>; bpf@ietf.org <bpf@ietf.org>; V6 Ops List <v6ops@ietf.org>

Cc: chinmayasharma.201ec113@nitk.edu.in <chinmayasharma.201ec113@nitk.edu.in>; tahiliani@nitk.edu.in <tahiliani@nitk.edu.in>; amoghumesh.201cs207@nitk.edu.in <amoghumesh.201cs207@nitk.edu.in>; balajinaidu.191ec111@nitk.edu.in <balajinaidu.191ec111@nitk.edu.in>

Sent: Tuesday, February 27, 2024 at 12:58:45 AM PST

Subject: RE: New Version Notification for draft-elkins-v6ops-bpf-pdm-ebpf-00.txt

Interesting and useful draft Nalini, thanks for forwarding to the list. 

We could use more of these types of implementation drafts particularly for EH. Hopefully you can get some time to discuss this in Brisbane. This may also be of interest to computing-aware traffic steering (cats) wg.

We also just updated the eh-use-cases draft which also includes PDM.

Thanks,

mike

From: nalini.elkins@insidethestack.com <nalini.elkins@insidethestack.com>

To: bpf@ietf.org <bpf@ietf.org>; V6ops List <v6ops@ietf.org>

Cc: CHINMAYA SHARMA . <chinmayasharma.201ec113@nitk.edu.in>; Mohit P. Tahiliani <tahiliani@nitk.edu.in>; AMOGH UMESH <amoghumesh.201cs207@nitk.edu.in>; Balajinaidu V. <balajinaidu.191ec111@nitk.edu.in>

Sent: Tuesday, February 20, 2024 at 05:49:55 AM PST

Subject: Fw: New Version Notification for draft-elkins-v6ops-bpf-pdm-ebpf-00.txt

All,

We have described our implementation of the IPv6 Destination Options extension header using eBPF.   We have also provided a performance analysis of the eBPF implementation in comparison to the traditional kernel implementation.

Please do read and comment.  We hope this may be of use to others in using IPv6 extension headers as well as to the BPF community in looking at the performance analysis of various factors such as CPU, latency, throughput and so on.

Thanks,

Nalini Elkins
CEO and Founder
Inside Products, Inc.

https://www.insidethestack.com

President

Industry Network Technology Council

https://www.industrynetcouncil.org

----- Forwarded Message -----

From: "internet-drafts@ietf.org" <internet-drafts@ietf.org>

To: Mohit P. Tahiliani <tahiliani@nitk.edu.in>; Amogh Umesh <amoghumesh02@gmail.com>; Balajinaidu V <balajinaiduhanur@gmail.com>; Chinmaya Sharma <chinmaysharma1020@gmail.com>; Mohit Tahiliani <tahiliani@nitk.edu.in>; Nalini Elkins <nalini.elkins@insidethestack.com>

Sent: Tuesday, February 20, 2024 at 05:45:33 AM PST

Subject: New Version Notification for draft-elkins-v6ops-bpf-pdm-ebpf-00.txt

A new version of Internet-Draft draft-elkins-v6ops-bpf-pdm-ebpf-00.txt has

been successfully submitted by Nalini Elkins and posted to the

IETF repository.

Name:    draft-elkins-v6ops-bpf-pdm-ebpf

Revision: 00

Title:    Implementation and Performance Evaluation of PDM using eBPF

Date:    2024-02-20

Group:    Individual Submission

Pages:    16

URL:       https://www.ietf.org/archive/id/draft-elkins-v6ops-bpf-pdm-ebpf-00.txt

Status:   https://datatracker.ietf.org/doc/draft-elkins-v6ops-bpf-pdm-ebpf/

HTMLized:  https://datatracker.ietf.org/doc/html/draft-elkins-v6ops-bpf-pdm-ebpf

Abstract:

  RFC8250 describes an optional Destination Option (DO) header embedded

  in each packet to provide sequence numbers and timing information as

  a basis for measurements.  As kernel implementation can be complex

  and time-consuming, this document describes the implementation of the

  Performance and Diagnostic Metrics (PDM) extension header using eBPF

  in the Linux kernel's Traffic Control (TC) subsystem.  The document

  also provides a performance analysis of the eBPF implementation in

  comparison to the traditional kernel implementation.

The IETF Secretariat

-- 
Cats mailing list
Cats@ietf.org
https://www.ietf.org/mailman/listinfo/cats