Minutes IETF106: rift
minutes-106-rift-01

RIFT minutes:

WG status:
•       Alvaro Retana: Since this was through our LC. But how are we listen
what should we do with the shepherd. The authors, how are we reviewing the
model itself. And then, doesn't make them a question. I noticed that we talked
about modeling first which is fine. But since we're the first ones to do it.
We're probably going to have to go through any growing pains. So I noticed some
things, the attendance is not that great, which means that you're fine but we
should make sure that the model is reviewed. Somehow, I don't know that we have
anyone who beyond the authors who know threshed. They don't have a doctor or
doctors. I noticed that the reference, that's in there is public which is good.
And I don't know that it's stable, in the sense that it points to the LDL, I
think again. And when you look at that page it's like well this is for sure
that, because I just don't know if that makes a difference, going forward. If a
different version is going to be a different result that's going to affect
inter operability and feature or not. And if you have a preference of each say
well enough for us to look at, which goes back to your reviewing the sentence
self respect that it compiles runs and multiple relations doesn't really mean
that. Historically, I mean, we go through this all the time in the YANG model,
so the island. Hey, but then the doctor say well you should really promote it
this way, not so much a question, but I think we need to think about this stuff.

•       Jeff Tantsura: Very valued concern. I tried to find people outside of
IETF who knows RIFT. Since we are iterate the YANG model. After RIFT published.
And if you found by building new implementation. If something doesn't work we
can make a bis.

•       Tony Przygienda: finding someone review the draft is more up to the
chairs. There's tons of industry on focus on you know in our field. Now, why
should they have the chair for like the salesman's job. And so the RIFT
compiler. Well, we will be facing the same problems from GRPC2 to GRPC3, right,
yeah, good luck to you. RIFT has been very very stable. You know, put this
stuff, way back, and the compliance’s stable. There is no like official binding
spec of RIFT language but need software GRPC. And I wish we had the YANG that
does the job that we don't. And maybe we'll have long term. So that's about the
best answer that's going to be given right now.

•       Alvaro Retana: But so I don't know if the IETF has done any formal
specifications you seem to say GRPC. So if you haven't done. It doesn't matter.
But every other modeling language that we have doctor to look at. right now you
said if Apache is going to change something that you said, there may be issues.

•       Tony Przygienda: So before I jump into the groove of the talk. So,
first thing, it's more subtle, because the compound is deliberate actually
decoding different target languages, uh, issues between languages. Welcome to
the beautiful world of modeling. That's why I'm so happy we have actually three
programming languages and implementations and each of them have different
mechanisms. And they are implemented and all interior. Second, we very
carefully check but RIFT is promised we all meet our mentor the optional
without the default. And we can we can decode something we added an element and
so on. So we need a break me out say mo champion with Bruno for this whole
time, because otherwise we wouldn't progress the model without like doing go go
even the major versions. But yes, I mean the street is doing tons of ideal now,
it's all over the place and we kind of have to catch up and I guess over time
we'll probably have to involve in the YANG we have the energy and the time to
do those things, otherwise we have to learn to live with all these ideal.

•       Bruno Rijsman (Jeff Tantsura read): I did run into some issues where
the Thrift compiler for one language (e.g. Python) behaved somewhat differently
than another language (Rust). E.g. handling of unnumbered.

•       Alvaro Retana: Well we're supposed to be producing interoperable
specifications right.

•       Jeff Tantsura: Yes, incomparable of the compiler behavior.

•       Alvaro Retana: We should at least include. I just put the draft and
there's something. There's an implementation considerations section. Maybe you
want to do implementations status section that talks about a little bit already
done. Yeah, this compiled in this language. So that when people look at this.
At least you know they might not have the same. There's an RFC that talks about
best practices in positions to basically a section that talks about the current
state. And in fact, it is removed. When the RFC is published. This doesn’t mean
the objections.

•       Jeff Tantsura: We will discuss this.

•       Tony Przygienda: So yeah. Okay, no problem. It’s a new section. So the
notification section is basically up before the stuff gets you know RFC, as
proposed standard. As a method observation. This is no different than having
the implementation and C compiler bar. just a different level. By the end, it's
not like we fundamentally changing the world right we're getting better tools
and each of the tools may affect that the resulting output and it will be fixed
by interested people around this stuff. It's not fundamentally it moving target.

•       Alvaro Retana: Yeah. I agree. just, you know, trying to guess what
issues come later, so that we try and talk about and not wait until the
registry right.

•       Jeff Tantsura: Fortunately, no one offends the status.

1. Dmitry. Afanasiev presenting multi-homing and controlled disaggregation
•       no comments.

2. Tony Przygienda presenting RIFT UPDATE

•       Tony Przygienda: should we put something (BFD) into the schema?

•       Jeff Tantsura: if you make it normative you'll have to wait.

•       Tony Przygienda: I don’t know if the BFD strict mode should be
normative. so like you can safely ignore it everything will work but if you
don't have to indication you don't know whether the other guy desires and
strict so like you have no signaling you you may ignore the signal by you don't
have to signal you cannot know whether you're willing to use it or ignore.

•       Alvaro Retana: it just become and the fact that it's optional doesn't
mean that it doesn't need to be normal.

•       Jeff Tantsura: so it's a separate draft on strict BD.

•       Tony Przygienda: yeah but you unique the information element to
indicate that you want it strict which would basically necessitate a minor
schema revision.

•       Bruno Rijsman (Jeff Tantsura read): the registry is not the right place
to publish a model we don't public YANG models in the registry.

•       Tony Przygienda: do yeah so here's observation Bruno is always almost
right he can talk and that's good so there is a big value to the registry I see
and you will see that when so let's say we have this schema version two and now
we go to version 3 so like yeah okay.

•       Jeff Tantsura: I learn what do you think I mean it starts a discussion
years ago right how we're going to normatively what are we doing with the
schema.

•       Alvaro Retana: so the document right now has a long a section I'm
assuming that you're doing registers right now.

•       Tony Przygienda: right well which is a news item generated I compile
the schema and I produce this wonderful IANA section.

•       Alvaro Retana: but it's all the same reason that you just said avoid
collisions and in case of extensions something that I agree I have no idea how
often major schema verses are gonna come by yeah or anything like that because
if it's every six months then who cares I just engine stand I think things are
really make sense.

•       Tony Przygienda: so either this stop will be not successful which means
quiet life or otherwise if this thing becomes let's assume it becomes wildly
successful just like is ISIS then I mean what is the resolution I mean how
often do we extend ISIS.

•       Bruno Rijsman (Jeff Tantsura read): so I'm not Jeffrey so they say I'm
a green let's just serialize the models themselves here publishes Model to one
first wins the other party has to fix the merge conflict and publish to.

•       Tony Przygienda: yeah but then you know how do we publish models
because right now it's just part of a document I stuck it isn't append exist at
the right way or maybe we have a special registry type called schema I know.
but I mean we I think we won't be able to avoid the benefits of schema.

•       Pascal Thubert: the RFC editor is usually fantastic at helping doing
this I mean could we just open up a bit the envelope of what we ask from them
and say eh in this case you're allowed to look a bit more into the working of
our sentences because they help already doing this.

•       Alvaro Retana: we need document to be readable and reviewable by a
bunch of other people so I would prefer if it's Jeff's just before if this was
done before in other words we don't want to just leave everything to the editor
because you know they have a lot of work done.

•       Tony Przygienda: I also think it's non-productive right because what
you want is sit there and have someone read and ask question and smooth it it's
not like you throw it over to the editor it comes with a bunch of question he's
like hungry 140 pages right so I mean I got these comments and it wasn't like
it's unusable it's more like well no it could be smoother all right sentences
sometimes too long edging on here get on the grammar look at stuff so no I need
volunteers who are willing to sit in a room no in a room for an hour and read
through sound as well that doesn't read well can you like smoothie I mean I'm
not I'm not I'm doing the editing so not a big deal but it just needs someone
who will put the brain to look you through another lens

•       Bruno Rijsman (Jeff Tantsura read): I suggest just given up on going
SVG diagrams to support that's not good enough it is a huge hurdle in reading
text version let's just do ASCII art fsm diagram instead I can do it not really
but next week.

•       Tony Przygienda: sure I mean these things are complex so I don't know
whether we can squeeze in ASCII well I assume Bruno can or we can break it up
into multiple HSV's just subsets of transitions publish these things are so
bloody rigid you touch anything you know you have to manually update it but
yeah so I don't know how good the SVG stuff I just started and I saw that it's
pretty arcane but I assume it works if it doesn't and of course you know it's
much less fun.

•       Alvaro Retana: format its document already in version 3 or just
relationship to version 2.

•       Tony Przygienda: no I started on version 3 and I look for example I
realized this one RFC published in version 3 right the one and he doesn't have
any picture so like yeah yeah exactly.

•       John Scudder: it occurs to me that since you spent so much time
intimidating people by mentioning the page count on your document and then
saying that you want detailed review

•       Tony Przygienda: lots of these appendix don't worry.

•       John Scudder: and since you're so good at parallelism maybe you know it
might work I mean it's you get a certain quality of review this way but to the
extent that what you're trying to do is smooth out sentence as it could work as
you could farm out you know sections to different reviewers so maybe I don't
have time to do 150 pages for you but maybe 20 pages would be pretty easy.

•       Tony Przygienda: to a certain extent I think the introductory I mean so
I have been misled by this group who was originally very you know insistent
that they like in their ativ rather than the drawers back and then their
ratings do not paralyse that well but you know they're the narrative' is the
front so I don't think that can be practically split up especially it has to
like lead one into the next you have to understand why you disaggregate and you
understand what is a Fallen Leaf and why you need negative but from their own
yeah like ZTP could be probably just completely ripped out and reviewed
separately

•       Jeff Tantsura: John have you just volunteered to to review it?

•       John Scudder: for the purposes of the minutes I have absolutely not
agreed to do that. (Joke)

3. Dmitry. Afanasiev presenting draft-wei-rift-applicability

•       Pascal Thubert: so I understand for the anycast sometimes you don't
want to route down within the pod you want to go to the other pod.

•       Dmitry. Afanasiev: yeah we want to go to all the destination and
distribute traffic more or less equally if you have some anycast destination
which is very close to us basically try and normally traffic routes shortcut to
it

•       Pascal Thubert: yeah and that's because we actually put in the rib the
the information from the pod itself and we had the discussion about tunneling a
long time ago and maybe you will resurrect those ideas in a separate draft but
there might be cases where are you in the in the top of that you don't want you
learn this thing from the flooding but you actually don't put it in your
routing table you just pass it up.

•       Dmitry. Afanasiev: sort of.

•       Pascal Thubert: if you do that then you get the kind of results you
want right because because you don't have the route self to the unicast address
there are cases for me the case was more mobility if things move very rapidly
you might want send them to the tof because maybe I need a tof as a real state
of where things are.

•       Dmitry. Afanasiev: yeah that's basically because we have ties which
prescribe to which direction we can bind the route if it were talking about
anything not using some direction for any go to the top of fabric and then go
other option is essentially to say in designs and network around so we have
this unicast destinations not in the same ports where very generate traffic.

•       Tony Przygienda: otherwise you know we can have something I mean we
have communities on the prefixes right so you can put the community on the
prefix do not install in the RIB unless you're tough we can just reserve a
community. does a very simple solution right?

•       Pascal Thubert:  make sense.

•       Dmitry. Afanasiev: anycast is when a real problem in data center
fabrics.

4. Tony Przygienda presenting additional applicability pictures

•       Dmitry. Afanasiev: when we inject in specific are basically horse
trout's on the back routes for top of fabric notes and propagating them down
it's much easier on the hardware resources because all those routes are
essentially pass it's they're not consuming it simply resources so it's
certainly not difficult to handle.

•       Tony Przygienda: well not necessarily an order yeah depends also new
implementation how you pick it out right I mean but because if if beta
advertises loop I get slots it to everyone right so these well these guys is
unicast but if it floods farther down that the these guys will have multiple
paths

•       Dmitry. Afanasiev: yes I said it depends on the topology in this one we
have additional meshing between spine levels so yes there will be multiple
paths. anyway it's not going to be very wide at least.

•       Tony Przygienda: well is as wide as you fan out in the port. right yeah
okay yeah also that you concerned it would be like port file.