[P4-arch] P4 TM SubWG Meeting 2 Minutes

Andy Fingerhut andy.fingerhut at gmail.com
Fri Nov 2 11:53:17 EDT 2018

I have added links to these P4 code examples for RED and WRED in the
meeting minutes Google Doc that Steve linked to in the earlier email, but
putting them in an email to p4-arch so everyone who is interested will be
notified of their existence.  Comments/corrections/questions welcome.  They
are pretty straightforward.




On Thu, Nov 1, 2018 at 3:12 PM Steve Ibanez <sibanez at stanford.edu> wrote:

> Hi Everyone,
> I've added the meeting minutes and slides into the google drive folder [1].
> Some food for thought:
>    - It seems like a key extern that is currently missing from P4 is one
>    that computes time averages. The P4 program provides samples and the extern
>    returns the average of the current and previous samples over some
>    configured time interval.
>    - The averaging interval would be configured by the control plane. If
>    the interval is over a longer period of time then the P4 program would be
>    able to extract fairly stable flow rate estimates (in fact this would
>    probably replace meter externs). If the averaging interval is smaller then
>    the P4 program could accurately perform microburst detection
>    <https://www.cs.princeton.edu/~jrex/papers/snappy18.pdf>.
>    - This would also allow the buffer extern that we discussed today to
>    expose only instantaneous queue size values and the P4 program can turn
>    those samples into an average if desired.
> Best,
> Steve
> [1] https://drive.google.com/open?id=1oys0EJUC_UR9vTCRBFCVBEFH4ka5x8af
> _______________________________________________
> P4-arch mailing list
> P4-arch at lists.p4.org
> http://lists.p4.org/mailman/listinfo/p4-arch_lists.p4.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.p4.org/pipermail/p4-arch_lists.p4.org/attachments/20181102/0fc730e8/attachment.html>

More information about the P4-arch mailing list