Re: [Dart] WGLC: draft-ietf-dart-dscp-rtp-02

Hi Brian,

with EF I link an expectation of excess traffic being 
dropped.

A rate scheduler with single digit millisecond buffer 
depth and tail drop like behavior is not an issue. 
To me, that's a shaper configured to be EF compliant. 
If such a constraint is not desired then the text 
may say "if mapped to a non-EF PHB, excess traffic 
may be shaped". I'd like to ask for some explaining 
text if shaping of EF excess traffic is regarded 
as a property to be expected for this PHB.

If an AF class is used to transport EF, then any potential 
remarking should be done in a way avoiding re-ordering 
(i.e. remark to DSCPs within the same AF class only). 
There's no point in remarking excess EF traffic to a PHB 
which may cause reordering. I think that's what a text 
on remarking of EF should express, if remarking is 
kept as an option to be expected (I've not yet heard of 
such a deployment).

Regards,

Ruediger

-----Ursprüngliche Nachricht-----
Von: Brian E Carpenter [mailto:brian.e.carpenter@gmail.com] 
Gesendet: Montag, 11. August 2014 22:13
An: Geib, Rüdiger
Cc: david.black@emc.com; dart@ietf.org; tsvwg@ietf.org
Betreff: Re: [Dart] WGLC: draft-ietf-dart-dscp-rtp-02

Hi Ruediger,

On 11/08/2014 22:21, Ruediger.Geib@telekom.de wrote:
> Hi David,
> 
> new EF related text has been added in the -01 version. I've copied in 
> some text of RFC3246 related to EF remarking and policing below.
> My take:
> 
> . remarking of entire flows is an option, but no remarking of 
>   packets exceeding EF capacity limits.
> - looking at the EF jitter and delay properties, I doubt that 
>   shaping is desirable. RFC3246 doesn't mention shaping as part 
>   of EF conditioning. I personally wouldn't expect EF shaping 
>   in sections where policing is applied. 
> - dropping of packets exceeding EF capacity is a MUST, if I 
>   interpret RFC3246 correctly.

Well, it seems to me to be a SHOULD, with the MUST that you cite below being conditional on "a mechanism that allows unlimited preemption of other traffic." So I would not be at all surprised by deployments that support shaping and remarking. Also, there are certainly implementations that are closer to RFC 2598 or RFC 3248 and therefore less strict.

I think the draft should make it clear that there is likely to be variation in actual behaviour for EF traffic that exceeds the nominal rate. (Leaving aside the fact that without a string of matching SLAs along the path, the nominal rate might not even be well defined.)

   Brian

> 
> I suggest to adapt the section of draft-dart-dscp to be more precise 
> regarding the optional remarking of EF traffic and to remove shaping 
> as a conditioning option if traffic is beyond the EF forwarding 
> capacity limit. A Proposal for the final
> sentence:
> 
> ...exceeds that capacity must be dropped. EF compliant flows may be 
> remarked to a different DSCP.
> 
> All other new text indicated by the Diff version looks allright.
> 
> Regards,
> 
> Ruediger
> 
> 
> 
> draft-dart-dscp:
> 
> 3. Expedited Forwarding (EF) [RFC3246] intended for inelastic	
>    traffic. Beyond the basic EF PHB, the VOICE-ADMIT PHB [RFC5865]	
>    is an admission controlled variant of the EF PHB. Both of these	
>    PHBs are based on pre-configured limited forwarding capacity;	
>    traffic that exceeds that capacity may be shaped, remarked to a	
>    different DSCP, or dropped.
> 
> ########################################
> 
> RFC3246 (three excerpts):
> 
>    Packets belonging to a single microflow within the EF aggregate
>    passing through a device SHOULD NOT experience re-ordering in normal
>    operation of the device.
> 
>    Packets marked for EF PHB MAY be remarked at a DS domain boundary
>    only to other codepoints that satisfy the EF PHB.  Packets marked for
>    EF PHBs SHOULD NOT be demoted or promoted to another PHB by a DS
>    domain.
> 
>    If the EF PHB is implemented by a mechanism that allows unlimited
>    preemption of other traffic (e.g., a priority queue), the
>    implementation MUST include some means to limit the damage EF traffic
>    could inflict on other traffic (e.g., a token bucket rate limiter).
>    Traffic that exceeds this limit MUST be discarded.
> 
> 
> 
> 
> 
> -----Ursprüngliche Nachricht-----
> Von: Dart [mailto:dart-bounces@ietf.org] Im Auftrag von Ben Campbell
> Gesendet: Samstag, 9. August 2014 19:50
> An: dart@ietf.org
> Cc: Richard Barnes; draft-ietf-dart-dscp-rtp.all@tools.ietf.org; 
> mls.ietf@gmail.com
> Betreff: [Dart] WGLC: draft-ietf-dart-dscp-rtp-02
> 
> This is a DART Working Group Last Call of draft-ietf-dart-dscp-rtp-02. It's available on the data tracker at the following URL:
> 
> https://datatracker.ietf.org/doc/draft-ietf-dart-dscp-rtp/
> 
> The WGLC will conclude on 22 August, 2014. Please send your comments to the authors and the DART mailing list. If you've reviewed it and think it's good to go, please say so.
> 
> Thanks!
> 
> Ben.
> 
> _______________________________________________
> Dart mailing list
> Dart@ietf.org
> https://www.ietf.org/mailman/listinfo/dart
> 
> _______________________________________________
> Dart mailing list
> Dart@ietf.org
> https://www.ietf.org/mailman/listinfo/dart
> .
> 