Re: [tsvwg] Classifying on ECT(1) (was: Adoption call for draft-white-tsvwg-l4sops - to conclude 24th March 2021)

[Sebastian Moeller <[e-mail]>]

Hi Ruediger,


thank you very much, anybody else here knowing the "cost" of FPM, please let us know.

Best Regards
	Sebastian


> On Mar 24, 2021, at 12:04, <[e-mail]> <[e-mail]> wrote:
> 
> Hi Sebastian,
> 
> I've heard about bitmask filtering, but I don't deploy that. So far, making sure that the DSCP values of incoming traffic don't cause harm is part of my job, but that doesn't include the ECN bits. So I can't judge the impact and convenience of FPM filtering.
> 
> Regards,
> 
> Ruediger
> 
> 
> 
> -----Ursprüngliche Nachricht-----
> Von: Sebastian Moeller <[e-mail]> 
> Gesendet: Mittwoch, 24. März 2021 10:26
> An: Geib, Rüdiger <[e-mail]>
> Cc: [e-mail]; [e-mail]
> Betreff: Re: [tsvwg] Classifying on ECT(1) (was: Adoption call for draft-white-tsvwg-l4sops - to conclude 24th March 2021)
> 
> Hi Ruediger,
> 
> 
> do you know the cost of doing FPM filtering on all packets? That is, is such filtering something operators do routinely in production networks (as I expect), or is that a feature used for debugging mostly due to high computational cost? 
> 
> 
> Best Regards
> 	Sebastian
> 
> 
>> On Mar 24, 2021, at 07:56, <[e-mail]> <[e-mail]> wrote:
>> 
>> Hi Steve,
>> 
>> I'd suggest to concentrate on routers present at interconnection / IP-layer (if an operator doesn't participate in an L4S experiment and you assume presence of L4S traffic, peering and transit interconnections likely act as sources, I assume). Juniper PTX is designed as a Label Switch Router. It can be deployed on IP layer too, but as far as I recall, its capabilities are limited as compared to Juniper MX.
>> 
>> Regards, Ruediger
>> 
>> -----Ursprüngliche Nachricht-----
>> Von: tsvwg <[e-mail]> Im Auftrag von Steven Blake
>> Gesendet: Mittwoch, 24. März 2021 01:07
>> An: Tilmans, Olivier (Nokia - BE/Antwerp) 
>> <[e-mail]>
>> Cc: [e-mail]
>> Betreff: Re: [tsvwg] Classifying on ECT(1) (was: Adoption call for 
>> draft-white-tsvwg-l4sops - to conclude 24th March 2021)
>> 
>> On Tue, 2021-03-23 at 10:06 +0000, Tilmans, Olivier (Nokia -
>> BE/Antwerp) wrote:
>>> Hi,
>>> 
>>>> Sec. 4 (Operator of a Network) of the draft presumes that deployed 
>>>> equipment is capable to classifying packets specifically on ECT(1).
>>>> Have the authors confirmed that this feature is available on
>>> commonly  > deployed operator gear (e.g., IOS-XR, JUNOS)?
>>> 
>>> FPM on IOS/flexible-match firewall filters on Junos enable you to 
>>> classify  based on arbitrary bit masks, and use classification 
>>> results as usual, e.g., map to qos-groups/CoS/VRF/...
>>> 
>>> The associated deployment complexity/feasibility will of course vary 
>>> on a config by config basis, i.e., your NOC is likely your best 
>>> source of answers.
>>> 
>>> 
>>> Best,
>>> Olivier
>> 
>> Thanks. Are you sure FPM is supported on IOS-XR? I'm not finding any evidence of that. I'm not finding any evidence that it is supported on Juniper PTX boxes, either, and the story for MX boxes seems confusing.
>> 
>> Since this is critical to enabling operators (not participating in an
>> experiment) to protect themselves from L4S traffic, it would be worthwhile to confirm this capability is widely available on deployed gear. Perhaps the draft could include some configuration information in an appendix?
>> 
>> 
>> Regards,
>> 
>> // Steve
>> 
>> 
>> 
>> 
>