Re: [ieee-ietf-coord] routing area design team on dataplane encapsulation considerations

[Alia Atlas <[e-mail]>]

Eric,

Thanks for helping clarify this to others!
I'll be more careful in phrasing that of course this is at the IP or MPLS
layer.
The WGs or BoFs mentioned (NVO3, SFC, BIER) are all doing that.

Alia

On Wed, Dec 10, 2014 at 6:18 PM, Eric Gray <[e-mail]> wrote:

>  Alia,
>
>
>
>                 I think it may be under control now.  J
>
>
>
> The issue was that – when someone with a different background (with
>
> perhaps less background in IP) sees a sweeping statement along the lines
> of –
>
>
>
>   “where we have different encapsulations for different purposes carrying
>
>    different data, each such encapsulation doesn't have to reinvent the
> wheel
>
>    for the above common issues.”
>
>
>
> - they start to feel slightly territorial/defensive.
>
>
>
>                 Looked at from an outsider’s perspective, it is not hard
> to see how this may
>
> be interpreted to mean “we’ve solved a bunch of problems and want to
> suggest
>
> that you use our solution when solving similar problems.”
>
>
>
>                 The subtlety in this case (at least for me) is that the
> intersection between the
>
> topics indicated and the ones that have been worked on (and, in some
> cases, been
>
> something of a problem) at multiple layers (read “multiple
> encapsulations”) is fairly
>
> substantial.
>
>
>
>                 As Pat said already, it certainly includes OAM, QoS and
> congestion avoidance.
>
> As someone else pointed out privately, it also includes multi-pathing
> (which is usually
>
> a touchy subject in any context), like ECMP and LAG.  And of course,
> security is also an
>
> area of significant overlap.
>
>
>
>                 The degree of coincidental congruency is extraordinary,
> but understandable
>
> when you consider that you’re bringing these issues up because they are
> always
>
> coming up in a similar context.
>
>
>
>                 In any case, when looked at as Brian pointed out, it is
> clear that none of this
>
> was included in what you were talking about.
>
>
>
>                 This is largely a matter of spin.
>
>
>
> I’ve tried to make that clearer to some others…
>
>
>
> --
>
> Eric
>
>
>
> *From:* Alia Atlas [mailto:[e-mail]]
> *Sent:* Wednesday, December 10, 2014 2:22 PM
> *To:* Eric Gray
> *Cc:* Brian Haberman; [e-mail]
> *Subject:* Re: [ieee-ietf-coord] routing area design team on dataplane
> encapsulation considerations
> *Importance:* High
>
>
>
> Eric,
>
>
>
> The intent is absolutely not to address DLL designers. It is to address
> the on-going work
>
> in the Routing Area for new encapsulations (NVO3, SFC, potentially BIER)
> and to provide
>
> advice and thought so that where things don't have to be done differently,
> they aren't.  For
>
> instance, we don't need three slightly different ways of handling OAM.
>
>
>
> The drafts that you alluded to are not a basis for the design team's
> work.  Pat Thaler was
>
> asked to be on the design team because of her experience in this area and
> interest in the NVO3 work.
>
>
>
> I'd be quite happy to discuss further and see what we can do to mitigate
> this misunderstanding.
>
>
>
> Regards,
>
> Alia
>
>
>
> On Wed, Dec 10, 2014 at 12:43 PM, Eric Gray <[e-mail]>
> wrote:
>
> Brian,
>
>         You are correct in terms of Alia's posting, though the distinction
> was a
> subtle one that had escaped me - possibly because of the similarity in
> wording
> of the advice draft, and the mention of specific DLL encapsulation
> technologies.
>
>         In this case, the draft I mentioned is likely unrelated to the
> work that
> is proposed for the design team.
>
>         It's also possible that I was reading into this some work that we
> might
> need to do (though I suspect that it would be a more natural fit for
> Transport,
> than for Routing) - as a generalization of the notions behind the advice
> draft.
>
>         In addition to the issues with cooperation between L2 and L3 layers
> for congestion management, there are likely to be similar issues with a
> number
> of the topics that Alia had listed (ECMP, for example, has similar issues).
>
>         I brought this up because a couple of IEEE 802.1 folks had read
> this as
> if it was suggesting that the IETF is looking to tell DLL designers how to
> design
> link layer encapsulations and protocols.
>
>         I see now that they were even more wrong than I thought.  :-)
>
>         Note, however, that we should at least consider whether or not we
> should track the advice draft, and/or similar work...
>
> --
> Eric
>
> -----Original Message-----
> From: ieee-ietf-coord [mailto:[e-mail]] On Behalf
> Of Brian Haberman
> Sent: Wednesday, December 10, 2014 12:26 PM
> To: [e-mail]
>
> Subject: Re: [ieee-ietf-coord] routing area design team on dataplane
> encapsulation considerations
>
> Hi Eric,
>
> On 12/10/14 11:53 AM, Eric Gray wrote:
> > Folks,
> >
> > Raising this issue in case it has not been raised already on this list.
> >
> > The issue is with an announcement made by one of the Routing ADs of
> > the appointment of a design team with intention to provide DLL
> > encapsulation and protocol designers with advice on common ways to
> > interact with IP encapsulation on a number of issues.
>
> I think the above is a mis-characterization of what Alia announced.  The
> design team is looking at data-plane, not data link layer, encapsulation
> within IETF protocols.  Think of protocols like MPLS and LISP or for that
> matter IP over UDP-based tunnels.
>
> Regards,
> Brian
>
> >
> > As I understand this, this is an effort that may have started with a
> > draft written in March of this year.  Pat Thaler is one of the 3
> > co-authors (with Bob Briscoe from BT and John Kaippallimalil from
> Huawei) on that draft.
> >
> > The current (-04) version of the above draft expired in September,
> > most likely because people wanted to figure out what to do about it
> > (and related work).
> >
> > That draft is:
> >     "Guidelines for Adding Congestion Notification to Protocols that
> >       Encasulate IP"
> >
> > http://tools.ietf.org/html/draft-briscoe-tsvwg-ecn-encap-guidelines
> >
> > There is also the Deterministic Networking (DetNet) WG, which sits
> > somewhat squarely in the middle of the "QoS" aspect of IP transport
> > protocols).
> >
> > And there is the recent BIER work that may possibly have an impact on
> > IP transport over a few data link technologies.
> >
> > I assume there is also related work on ECMP entropy, packet size and
> > fragmentation, OAM, security/privacy, extensibility and IPv6 header
> > protection as indicated in Alia's mail (see below).
> >
> > Alia sets what is in my opinion the right tone (consistent with the
> > draft mentioned above) in saying that the intention is to provide
> > advice on possible common ways to deal with these issues at the
> > data-link layer and how they may interact with IP.
> >
> > As long as the design team stays with the spirit of that tone, it is
> > likely there will be issues for IEEE 802.
> >
> > I believe this needs to be on our issues list for IEEE/IETF
> > cooperation, and folks should keep an eye on it.
> >
> > In particular, some folks have expressed concerns that this might be
> > view as implying that the IETF is expanding its charter to include DLL
> > design.  I do not see that, but that best way to deal with this as a
> > perception issue is to track the activity.
> >
> > --
> > Eric
> >
> >
> > From: routing-discussion [mailto:[e-mail]]
> > On Behalf Of Alia Atlas
> > Sent: Tuesday, December 09, 2014 5:47 PM
> > To: [e-mail]<mailto:[e-mail]>
> > Subject: routing area design team on dataplane encapsulation
> > considerations
> >
> > I have chartered a Routing Area Design Team to work on data-plane
> encapsulation considerations.
> >
> > I've bcc'd nvo3, sfc, bier, and rtgwg as the most directly relevant.
> Please keep any conversation in one place on routing-discussion.
> >
> > Erik Nordmark has kindly agreed to lead this design team.  The members
> > of the design team are:
> >
> >   Albert Tian <[e-mail]<mailto:[e-mail]
> >>
> >   Erik Nordmark <[e-mail]<mailto:[e-mail]>>
> >   Jesse Gross <[e-mail]<mailto:[e-mail]>>
> >   Jon Hudson <[e-mail]<mailto:[e-mail]>>
> >   Larry Kreeger (kreeger) <[e-mail]<mailto:[e-mail]>>
> >   Pankaj Garg <[e-mail]<mailto:
> [e-mail]>>
> >   Pat Thaler <[e-mail]<mailto:[e-mail]>>
> >   Tom Herbert <[e-mail]<mailto:[e-mail]>>
> >
> > The mailing list,
> > [e-mail]<mailto:
> [e-mail]>, is closed but the archives are
> publicly available at:
> >
> > http://www.ietf.org/mail-archive/web/rtg-dt-encap-considerations/curre
> > nt/maillist.html
> >
> > The Design Team is chartered as follows:
> >
> > There have been multiple efforts over the years that have resulted in
> new or modified data plane behaviors involving encapsulations. That
> includes IETF efforts like MPLS, LISP, and TRILL but also industry efforts
> like Vxlan and NVGRE.  These collectively can be seen as a source of
> insight into the properties that data planes need to meet.  The IETF is
> currently working on potentially new encapsulations in NVO3 and SFC and
> considering working on BIER. In addition there is work on tunneling in the
> INT area.
> >
> > This is a short term design team chartered to collect and construct
> useful advice to parties working on new or modified data plane behaviors
> that include additional encapsulations.  The goal is for the group to
> document useful advice gathered from interacting with ongoing efforts.  An
> Internet Draft will be produced for IETF92 to capture that advice, which
> will be discussed in RTGWG.
> >
> > Data plane encapsulations face a set of common issues such as:
> >
> >   * How to provide entropy for ECMP
> >   * Issues around packet size and fragmentation/reassembly
> >   * OAM - what support is needed in an encapsulation format?
> >   * Security and privacy.
> >   * QoS
> >   * Congestion Considerations
> >   * IPv6 header protection (non-zero UDP checksum over IPv6 issue)
> >   * Extensibility - e.g., for evolving OAM, security, and/or congestion
> control
> >   * Layering of multiple encapsulations e.g., SFC over NVO3 over BIER
> >
> > The design team will provide advice on those issues. The intention is
> that even where we have different encapsulations for different purposes
> carrying different data, each such encapsulation doesn't have to reinvent
> the wheel for the above common issues.
> > The design team will look across the routing area in particular at SFC,
> NVO3 and BIER. It will not be involved in comparing or analyzing any
> particular encapsulation formats proposed in those WGs and BoFs but instead
> focus on common advice.
> >
> > Regards,
> > Alia
> >
> >
> >
> > _______________________________________________
> > ieee-ietf-coord mailing list
> > [e-mail]
> > https://www.ietf.org/mailman/listinfo/ieee-ietf-coord
> >
>
> _______________________________________________
> ieee-ietf-coord mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/ieee-ietf-coord
>
>
>