Re: [Detnet] Magnus Westerlund's Discuss on draft-ietf-detnet-mpls-over-udp-ip-07: (with DISCUSS)

["Black, David" <[e-mail]>]

Stewart,

> I am not sure I understand why any additional text is needed.
>
> We agree that Detnet fulfils the requirements to run with no c/s.

I'm not sure whether you and Balazs are aligned on what to do.  Anyhow ... attempting to explain ...

The additional text would (briefly) describe how DetNet/MPLS/UDP/IPv6 fulfills RFC 7510's requirements for not using the UDP checksum (NB: the UDP checksum considerations for IPv4 are very different and do not need this level of attention, see RFC 7510, Section 3 [not 3.1]).  The text might be a brief explanation of why most (all?) DetNet/MPLS/UDP/IPv6 implementations satisfy exception a or b in Section 3.1 of RFC 7510.

Thanks, --David

> -----Original Message-----
> From: Stewart Bryant <[e-mail]>
> Sent: Thursday, December 3, 2020 12:26 PM
> To: Black, David
> Cc: Stewart Bryant; Balázs Varga A; Magnus Westerlund; The IESG;
> [e-mail]; [e-mail]; [e-mail];
> [e-mail]
> Subject: Re: [Detnet] Magnus Westerlund's Discuss on draft-ietf-detnet-mpls-over-
> udp-ip-07: (with DISCUSS)
> 
> 
> [EXTERNAL EMAIL]
> 
> I am not sure I understand why any additional text is needed.
> 
> We agree that Detnet fulfils the requirements to run with no c/s, and adding a c/s is
> going to degrade performance since forwarders in routers cannot see the whole of
> the packet in the fast path and so using a c/s is actually going to degrade the detent
> performance in many cases by adding latency.
> 
> So if we are going add text I think we need to add a warning not to use the c/s
> unless you have special h/w to support its calculation automatically.
> 
> On the other hand we could without risk remain silent and leave this to the
> implementors & operators to agree since there is no interoperability issue.
> 
> - Stewart
> 
> > On 2 Dec 2020, at 22:40, Black, David <[e-mail]> wrote:
> >
> > Hi Balazs,
> >
> > Digging in a little deeper, I concur with Magnus's underlying concern that this
> draft ought to say something about UDP checksums with IPv6.  There's a useful
> starting point at the end of the Introduction (Section 1):
> >
> >   As specified in [RFC7510]: "MPLS-in-UDP MUST NOT be used over the
> >   general Internet, or over non-cooperating network operators, to carry
> >   traffic that is not congestion controlled."  This does apply to
> >   DetNet networks as this document focuses on solutions for networks
> >   that are under a single administrative control or within a closed
> >   group of administrative control.
> >
> > That suggests that the first two exceptions (a & b)  in Section 3.1 of RFC 7510
> (both of which involve single administrative control) are more likely to apply to
> DetNet than the third one (c, based on higher layer recovery and/or error
> tolerance).  It could be helpful to say that in an added paragraph on UDP checksums
> (for both v4 and v6) at the end of Section 4 in this draft.
> >
> > I would also suggest aligning the text quoted above (from the end of Section 1)
> with the text used in exceptions a. & b. in Section 3.1 of RFC 7510, as I think roughly
> the same scope is intended.  In particular, it appears to me that this draft's notion
> of "closed group of administrative control" would fall under the notion of "single
> administrative control" in RFC 7510 (FWIW, I'm an author of RFC 7510).
> >
> > The suggestion to add use/non-use of UDP checksum to the list of management
> and control information in Section 5 is a good idea - that addition ought to cite
> Section 3.1 of RFC 7510 for the conditions under which the UDP checksum may be
> disabled for IPv6 (per RFC 7510, UDP checksum for IPv6 "MUST be implemented"
> for MPLS-in-UDP).
> >
> > Thanks, --David
> >
> >> -----Original Message-----
> >> From: detnet <[e-mail]> On Behalf Of Balázs Varga A
> >> Sent: Wednesday, December 2, 2020 12:25 PM
> >> To: Magnus Westerlund; The IESG
> >> Cc: [e-mail]; [e-mail];
> >> draft-ietf-detnet-mpls-over-udp- [e-mail]; [e-mail]
> >> Subject: Re: [Detnet] Magnus Westerlund's Discuss on
> >> draft-ietf-detnet-mpls-over-
> >> udp-ip-07: (with DISCUSS)
> >>
> >>
> >> [EXTERNAL EMAIL]
> >>
> >> Hi,
> >>
> >> Chapter 5. of draft-ietf-detnet-mpls-over-udp-ip provides a
> >> _non-exhaustive_ list of control and management plane information.
> >> DetNet does not changes rules of rfc7510: if the exceptions listed in
> >> 3.1 of rfc7510 applies, then using zero checksum is allowed; otherwise not.
> >>
> >> A possible solution can be to add an additional information element
> >> to the list in chapter 5, which allow or not the usage of zero-checksum.
> >>
> >> Thanks & Cheers
> >> Bala'zs
> >>
> >>
> >> -----Original Message-----
> >> From: detnet <[e-mail]> On Behalf Of Magnus Westerlund
> >> via Datatracker
> >> Sent: Wednesday, December 2, 2020 4:47 PM
> >> To: The IESG <[e-mail]>
> >> Cc: [e-mail]; [e-mail];
> >> draft-ietf-detnet-mpls-over-udp- [e-mail]; [e-mail]
> >> Subject: [Detnet] Magnus Westerlund's Discuss on
> >> draft-ietf-detnet-mpls-over-udp-
> >> ip-07: (with DISCUSS)
> >>
> >> Magnus Westerlund has entered the following ballot position for
> >> draft-ietf-detnet-mpls-over-udp-ip-07: Discuss
> >>
> >> When responding, please keep the subject line intact and reply to all
> >> email addresses included in the To and CC lines. (Feel free to cut
> >> this introductory paragraph, however.)
> >>
> >>
> >> Please refer to
> >> https://www.ietf.org/iesg/statement/discuss-criteria.html
> >> for more information about IESG DISCUSS and COMMENT positions.
> >>
> >>
> >> The document, along with other ballot positions, can be found here:
> >> https://datatracker.ietf.org/doc/draft-ietf-detnet-mpls-over-udp-ip/
> >>
> >>
> >>
> >> ---------------------------------------------------------------------
> >> -
> >> DISCUSS:
> >> ---------------------------------------------------------------------
> >> -
> >>
> >> So there might be something missing here in regards to zero-checksum
> >> in UDP when using IPv6. So Section 3.1 in RFC 7510 discusses this for
> >> MPLS over UDP and have some considerations that needs to be done if
> >> one are intending to use zero checksum. To me it appears that DETNET
> >> flows can not be guaranteed to always fulfill these, and in case you
> >> think you can motivate it should probably be stated explicitly and
> >> normatively allow it. So if it can't be guaranteed to fulfill these
> >> requirements then the next question exists: Do the possibility to use zero-
> checksum for this flow become something the control plane needs to signal it?
> >>
> >>
> >>
> >>
> >>
> >> _______________________________________________
> >> detnet mailing list
> >> [e-mail]
> >> https://www.ietf.org/mailman/listinfo/detnet
> >>
> >> _______________________________________________
> >> detnet mailing list
> >> [e-mail]
> >> https://www.ietf.org/mailman/listinfo/detnet