Internet Engineering Task Force 97, 14 Nov 2016

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: IETF97-LIME-20161114-1330

Description

LIME meeting session at IETF97
2016/11/14 1330

A

How many bacon brothers.

A

Newspaper in a gravel truck and the person's of home game java start with a longer here, free.

A

A

We need folders of women.

A

A

What we get off the surface, the Hitler was under the coconut oil communicate and budget.

A

A

A

Keep your pack.

A

A

With your quarters, 35.

B

Ok, let's start up, my name is Ron.

C

Bonica and.

B

I'll be sharing this meeting of line if you're here for lime, that's great, if you are here for a warm place to read the email. Well, that's, okay, we're friendly to UM for the people online who can't hear I see there are six seven, eight, nine ten eleven people in the room, not counting myself, Carlos Pena, Taro's interiors, not attending this I ATF. In any event, let's start out and as always, we start out with the note well slide. This is the IETF IPR policy.

B

If you're not familiar with the policy, please do familiarize yourself with it bore very bad. Things could happen.

B

Ok, let's start with the chairs slides. um Well, actually, I should put up the agenda first. Well. Actually, the agenda is rather short, as the chair slides a update on the connection less or documents that an update on the connection oriented documents, so it could could be a very short meeting and then we can use the time that we don't use informal meeting in a face-to-face update session face-to-face editing session. So, let's leap into the chair of slides.

B

Our Charter is on this website and if we go back to our Charter, we had three deliverables yang data models and we turned out having um we divided it into two at first and three later to around connection less protocols, one around connection oriented.

B

We were also going to have an architecture document and an applicability document, but those both got subsumed into the data models because they they were both about a paragraph long. So, basically, our task is to finish three documents and those are the three documents that we currently have as working group items. All the other documents have been subsumed now. What is our status?

B

Well, the two connectionless documents in my opinion and that's why there's a question mark there, they're really close to being ready for working group last call we had two really productive into her meetings and I. Think they're, pretty close to cooked the connection. Oriented document isn't all that far behind it. We learned an awful lot doing the connection less and we kind of know how to how to attack that one now. But all the news isn't cheery um it's much later than you think.

B

This working group was charted two years ago to do three young models and the yang models were due over a year ago. The window of opportunity is passing when this working group was initially chartered. We thought that we were building a framework that was augmented and different working groups that did different oam mechanisms would augment it. Bfd would put some stuff in, and everybody who doesn't know IM model would put something in well, as it turns out we've taken so long that these working groups are going ahead without us as time slips.

B

So does our relevance so.

B

Here is the working group chairs bombshell, um we're going to work really hard and get the connectionless drafts to the iesg by the next IETF meeting the connection oriented draft by the meeting after that, and we will shut down at that meeting regardless of what kind of progress we made. So, if we're not making significant progress in the next two ITF, that's probably an indicator that um that we're not making significant progress and probably won't get there before before the window of opportunity closes anyhow. um Is there any any strong reaction to any of that.

D

Front bumpers, quick questionnaire, um I'm.

E

Sorry, your name again Frank.

D

Brokers, what's the rationale? So if you go one slide back mm-hmm and you said that the connectionless is me already, no I think the connection orientated is a little further along than the connection less yes,.

B

Yes right so it gets one more IETF and I.

D

Think if you go one step or.

F

D

Do believe that the connection or I think you got the numbers wrong, so they the connection oriented. It is a little further, along than the connection less whoops, my bad right.

B

C

B

D

This slide, if.

B

I meant to give.

D

The connectionless a little bit more time. Sorry.

B

I right I reverse them on this slide.

D

B

Connection less should get more time. Okay,.

C

Because we're on the site, not in.

B

Good shape, I'll fix that slide and post it again. Greg, yes,.

C

B

What does Greg.

C

Greg risky I think that what I wonder is that at some point we agreed that. Yes, we work on a paradigm packet, switching networks, connection-oriented connectionless OEM, but we understand that we want to have some more abstract model that will unify this to into some single model, which could be very thin, very light, but still will be very helpful, because that will ultimately the realization of our goal. As.

C

Abstract OEM model for the multi-layer environment, because in a multi-layer environment the transport usually will be a connection-oriented network and then overlay will probably be connectionless. Okay,.

B

This sounds like a reasonable thing to find something that you can abstract from between connectionless and connection oriented and put it on its own domain. That is what you're saying. No yes, abstract. What's common factor out, what's common I'd agree that might that might be a good thing to do. The question is: can it be done in two I ETFs, or will it take six idf's.

F

C

With enough thrust and pitbull I, so it all depends on participation of the group. If you.

B

Can make that pig fly I encourage you to do so and we'll all have barbecue okay,.

G

Binoculars I'm confused here, because okay, the goal of lime was to get one young model. Then we said: okay, it's impossible to get one yang model me to divide us into two connection, elestren oriented. Then we divide into three for the method which I agree with now we're telling with was correct that maybe you want to go back the abstraction which was the goal initially, but it proved to be not possible. So what am I missing there? Well.

B

um There is a question as to whether it's possible or not- and it's probably not appropriate for the chair to say it is or isn't that's the work of the working group. But what the chair can say is, if you can do it correctly and quickly show it to me and if you can't then two models, three models or maybe no models- is the right answer. Okay,.

G

Understood then we conclude, it wasn't possible. A thesis was my understanding by dividing into connection lesson oriented. hmm My conclusion was that the groove concluded it was not possible in abstraction, mm-hmm.

D

B

I believe that was the conclusion too um I see some of the people who were in favor of splitting it up and don't see any of them coming to the mic.

B

That's an invitation.

E

My name is Jim, actually I, think obviously bigger punch up into several job, I think, is based on discussion and the consumers who will follow these condenser. If we break these consistent by too much it all of this job into why, I think I was also kind of confused.

C

Ragnar, ski again, okay, a little bit history, their initial draft to one group came from the true model and it took some time to agree that the trio presents connection-oriented through OEM presents connection-oriented OEM paradigm, and then we didn't split it. We just designated. We clarified what it does and then started the work on connectionless, but at a time of the discussion are at least my recollection was that we agreed that, even though we're working on technology specifically a.m. our ultimate goal is to have an abstraction layer that unifies these two models. Okay,.

B

C

It's not about merging them because they cannot be merged. They are distinct because the paradigms of the network are distinct, but we agreed that it's well worth effort to have an abstraction layer that uses. Probably I don't know we, we call the test points and test functions that will hide specifics of their network paradigm. Let.

B

Me ask: let me propose something and ask for a show of hands from the people in the room. I challenge you Greg to produce some text that describes the abstraction layer and I don't mean an email. I mean yang Yang model that describes the abstraction layer hang on. If within four weeks you can have that done, and the working group salutes it. Then it's adopted and it goes in to our work path.

B

If it's not complete or the working group doesn't salute it, then it may be the right thing to do, but we just don't have the time to do it, because the window of opportunity is closing.

C

There enough, I just want to mention that, probably before berlin, I send over a proposal, and I got no response to it.

B

Could you send it again and this time with yang tet with yang, I.

C

Send with some yang prototype that basically used a reference fat, ok,.

B

I'd ask you to send that again today and that will start discussion of it. I'll.

C

Try to dig it up this week. Ok.

B

Then sometime this week is yeah the.

C

Week is kind of busy.

B

By the way, does everybody like that proposal is anyone paying attention? Ok and jeff has is always paying attention? No.

E

It's a channel actually I'm, not sure I follow the Greg's proposal. Actually you know for sale, OEM model, actually solid dude on top of the length of logic model network approaching model has all the abstract thought logic. So we already do such kind of attraction. So, if we'll grab a compose another active Rosie- and we can take a look like that- but the way already chaiyya chaiyya us to do that.

C

The thing is that again in.

C

Connection oriented and connectionless networks, the topology could be described by the same model, but OEM is different, because if you are connection-oriented, then you have to care about their miss connection. Defects which is not.

B

The case in connectionless.

C

Okay, so there are some distinct differences in models. That is why we discussed and agreed to have to yank models for OEM connection-oriented elfin connectionless at.

B

This point probably debating about it in abstractions won't help very much. If you could make the proposal in email, we will discuss the proposal on the list and today's November 14th on December 14th will have a cut off on the discussion either. We accept the proposal or not, and we can remember that because December 14th is Beethoven's birthday Benoit.

B

G

Nautilus, so you know if the obstruction would speak about is simply like the endpoints. It could be like it tight death which anyway, we have hit apology right. That's the first thing: I should go to your next slide. Please shut down by ITF, 99 I've been hearing so many times that those draft and yang models are almost ready for so many meetings at this state is important. The window of mention I strongly believe in to that.

G

If we don't deliver, it does not matter right, because the different OEM mechanism will have their own yang models and we are not relevant actually.

C

I want to clarify, and so we already have yank models that I believe I should not be impacted by lime models, but it bearable be beneficial for the lime to use them, and I can name rbd young model, a t1, p and model. I can name mpls, LSP, pink young model and mpls-tp OEM en model, so these models already exist and I believe that why model has to accommodate them rather than force these models to change.

B

So clearly, but the question now is given that these models exist, what value does lie mad ah I.

C

Believe that there is a value, and especially the value, because we want to create some abstraction and so because, for the same of VFD, BFD is working in connection oriented networks in mpls-tp as it works in connectionless.

B

C

In itni p.m. pls.

B

So I'm drawing a mental model of what we would have if we continued with all of these models plus com lime, there would be end models which overlap, which model a partially overlapping set of stuff. Now, the partially over lettings overlapping stuff would be modeled one way for one protocol, another way for another protocol, and yet a third way by yang in the abstract way, I'm scratching my head and wondering what the value.

C

B

C

But again, in my view, are four existing models. Why, more, though, can argument are the b of d and be done with it, because you know you don't have to redefine how you enable continually check-in connectionless network, because you already have VFD model so.

B

Really, what you're doing is turning lime on its head at one time. Lime was the common area. That was an abstraction that all the other protocols didn't have to do now: lime, augments and other protocols, with with something that they already have in.

C

My view, the lime is their model of saying or enabling operator to execute or instantiate certain om functionality. Okay. So basically, if you're in the network- and you want to say okay- we want to monitor continuity between a and B. Then are you go into more technical, our technology in paradigm, specific definition which is based or argumentative I tech, nology, specific solution? How you do it, whether it's a BFD or it's something.

B

C

1731 like and then in that environment, you provide the specific parameters of your protocol.

B

Okay, let's see I want to go a little more into the role of chairing and not the debater. So does anybody else have any other any other feelings on this.

E

This is channel actually based on collaboration from the greg. I feel I did with the contradiction you know for for the motor design, though so I guys, I feel confused.

B

Okay, I think probably the best thing we can do is ask Greg to proposes yang model, proposes changes to the yang model, will evaluate them by December 14 and either move forward with them or not, but in any event, we're still looking at an ietf 99 deadline.

B

Any other comments.

B

Okay, um now we will move on to the this time. I'll get them right. I, always confuse the connection lesson connection-oriented. We will go with the connection less update first and.

B

And I have the long one.

B

And that is certainly the wrong one.

B

I thought this would be easier to control.

F

Okay, let's do it this way.

B

B

H

Good afternoon, everyone, my name, is Michael from hallway and Saturn I will introduce lime, the seal and the CI modules update. First all modules update ah since the Alaska ETF meeting. We have a lot of work to pulses, see how module move forward. We have several internal meeting which also join other internal meeting, his pfizer PFE working group in this.

H

In this legend terminating, they discuss the CL and seal models, update and also discuss some questions, such as the relationship between the line model and another model existed defined by the proto working group, and we also have a several discussion in the email list. I include her. Why should we need to split eights model into two and another is the relationship with land? Another module also have some discussion about ultimate, measured and skill.

H

Among master fund rate, we are request, as a young actually have young doctor to review our work and gave us some comments to our Simonson. They can make our multi-pattern know who this work move forward.

H

Okay, this about is a cell modules update, so most of all a changes we modify this places. It's a model into 2184 CL modules face motive, neither is traveling master and we also okay. This issue is a razor since last night of meeting and Frank initiated discussions, email list, most people's part to explicit modeling, 22 and working group makes its decision. Then they do relative update coding, chooses to Syrian and discussion and second important changes way aid a section to describe the LA sale modules applicability.

H

In this section we propose to use both CL module and another modules specifying the protocol working group as basis, therefore soul. I module can provide its structure and another technology Pacific mother can provide. These are details. The argument align with this group in defending this technology perfect module and therefore we can address a line this narrows to to to try some land problems such as quickly drew found a luta fault location, our stage different layers, Tessa roids out.

H

We also do some small change in the CL module, such as they ate some description to introduce a relationship with Tessa at where I asked. As a topology motive and the way also modified supporting juice attributed to read only and finally, we remove some CL relative technology attributes such as air, three VPN and unhealthy p. Relative attributes.

H

Ok, it is a piece above some important you to discussion the first days a week. This has a large about the relationship between the line and another modules before days, the topic and Greg and run, and the chin have discuss about this relationship.

H

I think we have already associated in terminating in the society meeting and also in email discussion and in the PFT PFT working groups in terminating. We also introduce our proposal I think we make some consensus that the technology basic model can define that standalone, dr., standalone module and CMO do and this specific module can work together run as this leads to a dry salam. This narrows.

C

A

C

um Okay, um the current state of connectionless model does not address my comments that I send when there was a adoption call for the connectionless model. I believe that there are mentation of Technology specific models, and that includes VFD and LSP. Think in a connectionless methods model that has to be part of the model, not applicability, because it's a part of the yang construct. And if we say that we do have a comprehensive VFD model in this case that it has to argument.

B

Augmentation of VFD.

C

Model has to be part of connectionless or a model, not applicability statement. Saying oh, you may do that. No, if we're saying that you need to do augmentation, because otherwise you would not be able to instantiate. Vfd are continually.

H

C

H

Okay for this question, why we're going to do some updating this model, because I was seeing if we want to shows a generic model wave ounces attribute our structure and more abstract and the way don't one to defend a lot of parameters it can describe in this techno specific modules.

C

H

So perhaps we want to address a land problem of the urban didn't, learn independence, a problem. Perhaps we need to just provide a structure of commons product model. Just come, can show the common structure and the common attribute, and if you have, if you have neither talk about specific module, is it cool? You can provide another solution for the user, the user can flexible chose to use it kind of a specific module uses module and the land structure built a new module to address it. Lance narrow, ok, I have.

C

Two questions to that. First: if you don't want to be a tied to the specific mechanism to do continuity check, then why are connectionless? Oem model includes the container with a list of RFC 58, 8158, 8350, 84 and so on, and so on. So that's very specific to me. Write a second of all are if you say that there are other mechanisms to do continue to check in connectionless networks. I would be interested to learn, which you mean.

H

Ok first first question is thus far into theta shows as captivity. It's now the defense in detail. It can shows this device of ok, I cants policies choose and we know not defined it it more. We not defined more technology, talk no specific details, so this is a part of I think it is a part of abstract module. It can provide ability. It can provide the letter user know of what what what should I quote. Excuse me about it. It's.

C

Quite specific that your listing I be the single hub, multi-hop and Vav over mpls, LSP and Vav, or to the way I think that's a very specific you're, not saying oh I can do continua t check and how I do it you'll figure it out when you want to do it?

C

Okay, I think that that's that's the level of abstraction for the line is I, do continue to check fine.

H

Okay, we can discuss it later and I.

C

I think that that's that's the discussion that we had for not the first idea and I sent the comments in working group. Adoption call and it's still there and at the same time you say that augmentation of VFD model is a to technology. Specific I think that these two statements don't connect to me.

B

Look I think somebody want. You want to comment on the same issue. Yeah.

D

Question on the combination, so red would it be? Is the proposal to just get rid of these references and with that solve the problem, because then that's a fairly easy approach solve the problem. ah That would be only half solution. In my view, I think.

C

That the solution would be yes remove. This list of RFC's not only related to BFD, but orbs is because they are too cryptic. They require a person to know what this RFC is about and instead of that use the functionality. If we are doing, for example, LSP thing, then its on demand continua tea and connectivity check, you can do it overall, LSP on to the wire with ls meeting for the 379.

C

This probably now the better the reference, but we have a model for LSP thing or just IP pain if it's likely a network okay, but you don't have to reference particular RFC, you just say on demand, continually check and activity verification period, everything else technology-specific and if we have already model like in case of LSP being just augmented there. That's.

B

Actually, fairly easy change to make would be okay, yes,.

C

B

The same in the same, in the same, to do with.

C

The DFD, instead of listing rfcs that we view related, say, are proactive, continually monitoring and then argument BFD to their okay, okay,.

H

I'd, like to flip axis comments to my closer and we can discuss a user first. We can't do relative change or update.

I

Such as a common trouble, san Cisco, a great union of mental model- or you mean reporter greetings from one of those lives, I.

C

Usually see that you argument the model.

I

Okay, so you've meant it. That means that the line model in infinitely like what we have also imprint almost other models. Well, oh.

C

Well, otherwise, how would you instantiate BFD, if you don't implement the v-model.

I

Like the don't you can, do you don't have just click your fee, you have to invent all the other earlier models, something augmented by the line as well. So it feels the discussion of the Lima augmenting of the emoji specific, not all to see, but the wrong announcement, so I mean fully green dating site. That's what see.

C

I

C

Again, um I don't see that if, because again, the lime, in my view of the line model, if we're talking about the guy I, don't know about other examples, because, yes, we can talk. Actually we agreed that performance measurement is will be separate and actually that's something that I realized. That is not on our timeline. So what will happen to the line performance measurement model?

C

That will something that I think that we missed in the earlier discussion today. But if we talk about, for example, our existing LSB thing model and the d model I think that they have to be implemented. If you want to realize the potential of connectionless om ah CFM is not connectionless. Cfm requires your connection. You define your MA in the mag and it's different. It's a connection. Oriented model.

C

Action, it's a different question. I think that we haven't really got to do how lime relates to ethernet models and, namely for any f38, because ma 38 is a young model of fault management for the carrier, ethernet service. So how why model and MAF 38 correspond.

I

C

Then with then we can, then we can ignore CFM.

I

C

I'm again, uh we had this discussion and how who I'm work related to ethernet service a.m. and we haven't decided. We I, don't know whether we have a clear solution. I think that our implicit agreement was that why model is applicable to technologies developed at ITA, which includes in the connection oriented domain.

B

C

C

But that's again, if I misrepresenting, please correct me because that that's my that's my that's my recollection of our discussion. We had this discussion. Yes, but we I think that agreed that we only lime covers thrill and mpls-tp.

H

Okay for this cross for six to issue, I want to do this again and.

C

H

For a stir and the working group and another such as carefully working groups, a discussion, we can some concern, sir, and we think we don't need to and argue choir to imagine either technology argument a lie or cheat engine lab, but you defined is a kind of specific module and our way provide is a structure and to address lamis narrows. We can use it to motor together to dress the problem: okay, knocks on okay. This is another discussion about a scheme around my certain document matter in the sale module applicability.

H

We introduce a matter that we can augment the mattress sale module with techno specific group to get as a to get the new, mario and grassland program, and another master is proposed by night mode. The working group ice cream among it also can solve the problem or to make a different model work together. Yes, man, yeah oxford house screaming bound hi.

F

I'm Roger, Luca and I'm a co-author of this scheme, amount and draft in the net, not working group and I'm a bit worried that now that scheme amount becomes sort of fashionable term that people will start using it instead of normal augment that that we have been using for several years. So the idea is that scheme amount is useful in cases where you want to use the same yen. Module in different places in your data are key.

F

Like you f, you have a module that could work for a simple physical device and then you also won't use it in some kind of realization, like in routing instance, and things like that. So suddenly, the same data becomes somewhere berating inside the hierarchy and then, of course, if you, for example, one to augment this module eight something else, you have to specify a precise place so and you cannot do it at the same time for both for the top-level location and for some mounted location. So that's for this.

F

This is the use case for scheme amount. If you talk about like adding some technology specific knobs and things like that, then in my view of a normal augment should work. So as long as you really have only a single place, or you want to put this new stuff, then an augment should work. So I would recommend you to look at at augment.

F

First, as as soon as you realize that it doesn't work for for our use case, then you may want to look at the scheme among, but as far as I can tell, this could work and also I'm a bit confused about the use of grouping, which is what probably also mention is.

H

Really, but a real the group, he is some group in defunding days, technology specifically such as dirty and obvious and Tiaras LCP mpls-tp, as it offends some a lot of reels in groups. We don't want to redefine disease attribute in Thailand, The Contender module. Therefore we reuse is cooking. Is they are not defending the land motive? There's a different me today.

F

I, don't know the details just to warn you that these groupings, you don't work like object-oriented classes, for example, so that really you you you if you want to use a grouping, let's use a specific grouping from a specific module. So it's not possible like to to have some object, oriented inheritance, and so you just go for a grouping into it. Will it will be somehow magically filton middle stuff that that's really really harmful for the particular device so but I just yes,.

H

This is about this.

F

H

Fourth, our land applicability, for example. You want to extend the alarm to Jocelyn their Theory, my module. You can reuse a lot, some a group in defunded dirty and the land structure, tis.

C

F

A way proposal.

H

The means a flick, bility yeah, so.

F

My conclusion this moment is just go for an augment possible and if you have any problems and feel free to contact me, if you on.

C

To us, thank you thinking.

C

Bergner skiing in I, just want to stress is that, if argument statement that list explicitly the model that we argument in till I'm Liam model connectionless is not part of their model, then it would not be augmented.

C

So it cannot be part of all you can argument. No, it has to be part of the base model. The augmentation has to be part of the base model. If our operator doesn't want to use proactive, a continuity verification, then it's fine, but you cannot do augmentation on a fly on the wheel. Okay, so please understand that.

H

Okay, this is a purple Emily faced for land. We just provide this a common structure on the we can now that defined is some techno specific modules. You know which each other way, clarity to Colonel, aesthetic module need to defending the protocol, working group and kind of specified working group. Therefore, in the lime audio, we we just can provide the Sun, it's a land applicability or some solution to the reader or to the user for declare.

H

Okay, our line module can provide this article city's ability- you can do this, but we don't 12 effort and why is ok, I.

C

Think that again, I have to point the contradiction of your statement and the fact that model lists their numerous RFC's which are technologists specific, so either remove all of them out of your model and have no connection to their instantiation, because otherwise, if you don't have a way of use, existing methods which are protocols or am protocols be of D or anything else, then I don't see the value of line model at all. Okay, if you cannot do from Lyme obstruction to go into technology specific protocol, then what's the value I.

C

Believe that their value of lime is that you say I want to do this or am functionality, and that gets you into technology specific or if you want to see the operational state say, give me the operational state between a and B and then it either. Does you on demand, verification or post the state of a proactive session? And how does its technology specific?

C

But if you don't have this connection to their technology, to the specificity of your paradigm? What's the value of lime.

G

Binoculars, so let me ask the stupid questions and goes I'm slightly confused, so connection less and bfg, which one or comment which one.

H

For lanza kwon do we use the landline at this motive and also in holds a BFA module, so we imported the FD module pocket. Fine. Now.

C

G

H

Are you know, cell and module liability? We do some tennis, okay, this.

E

Is your, let me clarity it a bit. Actually, we in our face model with not extend from base pft model, we empower network topological model and import a metal instance model, but we don't actually import a PFD model here. We show this applicability, we tell you how to extend it base model. So we talk about the youth organizer, honest apply to the model exchange not apply to them pace the model.

G

All right so I'm hearing that the bfg model will go along alone anyway. Right, that's one thing, so we want to have a line model that will be augmented for OEM, specific technology right yeah.

H

G

So, which one do we have? We've got like a potentially sfc or beer or auntie one. So these are those guys are talking to in film trying to get them to linked to lime, while for BFD too late. This.

H

The idea trail working groups, they have trail OEM, llamado and augment seal motive. Yeah, alright,.

G

B

So commitment, let's see, if I understand the answer to your question. The ultimate value of lime depends on our ability to influence these other working groups to augment line. What.

G

This is what a.

B

Charmander Stan.

G

It was too late for BFD already is.

B

Now the next question is: are the other groups highly motivated to do so or are we working on an airplane that will never leave the ground.

G

That's the question: what is the value of lime or we're too late already, if we just have like a module with a container, a pointer to an oem specific yang module I mean all right. So what is the value right? A container telling okay BFD.

B

Anyone this may be a derail in question, but maybe we should think about that for just a while and think about whether it makes sense to continue work.

F

Project just one comment, so if the situation is that you have some modules at our or they ready some other working groups- and you want to take these modules and put them to some place in your hierarchy, then of course it may be a use case or false king among because in this case the extra stuff that you want to add needn't care about your your basic structure. So they needn't put any augments in there more you. So you can just take their modules and put it somewhere. There are some limitations.

F

What you can do and cannot do, but that's of course the option, but I really don't know the details. I would have to look into what you are doing and trying to achieve, because.

E

I cannot tell me we can discuss it more. Just.

G

To come in on that below again, so, let's remember that charter was about consistent configuration, reporting and presentation if we change based on what letter set right, the configuration it's too late for the FD they're going to use their own module they're, not going to use a hierarchy that we've got. Even if you do a scheme amount to what they had right. That's the first thing so for BFD, the config part is over.

G

We might be able to influence some other OEM mechanism, the new ones, sfc beer or maybe t1 and I- don't know whatever will come right. Maybe if you got a nice model, we could augment that now to come back to BFD. Then then, the only value of lime is what the potentially the command the the consistent reporting, because they wink of a scheme amount. That says- and this is the container, that the pond amount points for BFD and it's the same level of the different mechanisms.

G

This is the only added value of lime in that case, which is tiny, but it.

D

Right so I think we just acknowledges for certain things that this is Frank for certain things. The train has clearly left the station via fig on trolled on fine check. There are other trains waiting in the station beer, maybe sfc. Let's go try right and given that the work has progressed quite far, let's give it a try.

D

Let's go publish this stuff and then if people pick it up and it's good, if people don't pick it up and it wasn't good enough so- and you have some printed paper- I think it's worth while continuing and then getting it done and trying to get it done quickly. As opposed to are seeing the question, should we go and continue right now, given that you're so close to the finish line.

G

Velocity I got not run was that it seemed that it took us like so much time at ease in little hat ujin's. Try to summarize this right and- and you know we asked to have a wiki with the different OEM mechanism connection, let fashion oriented whether we influence those guys, whether in their interest to augment line, etc.

G

We have to go back to this, which was requested like three meetings ago right, because if, in the the column telling is their interests to augment line- and the answer is no from everybody then, except to improve the reporting, then the value is.

B

Jeff Jeff tell us your name: hi.

J

My name is Jeff, hey.

B

J

Benoit, specifically about the BFD point, BFD does allow for configuration of VFD through other means. No, that's for the reasons why we have that dependency across several other modules. We have groupings exported for time and configuration so, for example, ospf can enable the FD sessions line can do exactly the same thing. I think your broader question is what should the general design pattern be? 40 am like things so that a device specific module can be not only in control of its own fate, but allow for other, no external things.

J

By being an example to do its job and similarly, how can those device specific modules you know, provide information that is useful so that the layer, independent portion of mine can actually get its job done. The first one in terms of general configuration, the FT, is obviously a good example of that Elle's ping. I haven't looked at the module as an example, so it would be interesting to see if there's some way to provide similar groupings to allow for external components to configuration.

J

I don't have a strong opinion about the layer, independent portions of the operational state.

B

Okay, so I will try to summarize what we just heard here and that is that again, the window of opportunity seems to have gone for BFD and for trill we may get that window of opportunity for other oam mechanisms. So let's work fast and get this done so aren't you know it seems like a relevance has to do with our time to publication and I'm. Sorry, you name slept Frank, I'm, sure, ok, Frank um Frank mentioned we're so close to the finish line.

B

Ok, if we get to get to the finish line, it's worth it if we take another year to get to the finish line, but the train will clearly have left the station. So anyhow, with that any more comments on connection less.

B

Let's talk about the connection oriented draft yeah, one more chance.

E

That it would continue jolicoeur ouya in the window for the DMV and ensure all close absolute zero. Today, standing on sale model, hmm no.

B

For me ft it's closed, but maybe for some other. So let's, let's try hard to get get done by ITF 99 and maybe we'll be there in time to be relevant for some other away and not mechanism. I.

F

Don't come up in fact, if flying proves to be useful, then nothing prevents somebody from adapting the BFD model just be able.

F

Well, we had the situation already weather radar outing module that also try to provide some some kind of a framework that routing protocol would use, but we now have also some more use for me GP, for example, that just ignore this work and just they wrote bethe model that starts from the top, and so what we've been trying to do is to convince people that maybe this framework can be useful for them and of course it will take time. So I think it's it's not carved in stone.

F

So if you want, you can just adapt their module, use a different name and put the VFD stuff using an augment or something if it's useful for you. So that's another option: I I! Don't think that the fact that bfg working group is done means that you cannot use their work in some other way. I need I, think you're, real candidate, ok,.

B

Any more on connection less: let's move on to connection-oriented, then.

H

Ok, so conduction or n de tomate. Your way, don't you updates and society meeting, but we prepared some bozo to chat some problem, and but you know that we have to allowed to enter meeting and not have many time to update Ahmad trapped, ser felices data on budget. This proposal is the right day already prepared and if we can submit, we can submit new version. This is talked about as a host version.

H

You have similar to the say, a multi-way I do a lot of work to prove this more work forward. It's in terminating and request a young, dr dre voh.

H

Ok in this new proposal way need to solve several questions. First issues how to deal with Sam McGrady, some makati, maybe some techno specific attribute and it equivalent to the medina, so Shh maintain association, ID and it be funded by a cutie. It also be used in technol hf, sir NPR's TP technologies. So this term this attribute miss Sims, a technology fake. Our solution very simple way: Chester remove it from the biz motive and if the multi-user want you to burn a date, 10 Oakland Monsieur module. Eight eight. Eight isn't a requirement oxide.

H

Neither the second issue is how to defend the knee placed the Middle East in the original version. Limes sale module. The me please define what is simple. With only have one, it is only one leaves interface, but we check the standard standard document. We found that the maple a if one two completions and made me at least we need to two parameters among address letters in eville, but we also found that maybe you are a created ultimate today. Another need to expatiate operation cooperation. This parameter is sims technology-specific.

H

So our solution is similar to choose a major deal with mark ID. We just remove it and if the user, one sees how to build a sukkah element, okay,.

H

Okay, it's a third question. Mr. issue, it's in a way I already the introducing CL module is a relationship between to another modules.

H

Okay, this is a final question. There are some AOE arrows in this module, so in the previous version the sale module can pass the PM exam, but enough camp as a young editor, a truth check and the way this is arrows and now it can pass the full of boxes to choose, and this way are deserve a proposal to update yeah, okay. So next up with rod to prepare another version according to this meeting to this discussion and require working group, another welcome google escoll thinking. This is a sale, module yeah.

C

I bregma risky, okay, so several comments regarding the relationship between this model and mpls-tp om model. So, regrettably, I see this model as a tool model. It's not mpls-tp or a model because we do have I'm pell, st peoria model and I suggested to argument it. So the document has information, formational reference, not the reference lists, mpls-tp OEM yang model as informational reference, but it not referenced anywhere in the text so which is kind of odd.

C

It's a stale reference which is not used and again I believe that, because mpls-tp is another connection, oriented network and developed at ITF next to trill, if you don't want to replicate mpls-tp OEM model, you need to argument this model into this base model. Okay, so uh otherwise you just replicating and duplicating mpls-tp OEM yang model that already exists.

B

Isn't this exactly the same conversation we just had a minute ago and yes,.

C

Your vacuum again so again, if, if you, if we agree- or we think that mpls-tp OEM model left the station, then remove the references from this document and make it trill model hear it.

B

Anymore comments.

B

Okay, thank you very much. Thank you. I'm going to try very hard to come up with some next steps um and I see two big two big obstacles to getting getting there. One obstacle is the one of factoring out some things that are common between connection oriented and connection less stuff, and here we have come up with the action item that Greg mirsky will generate some text by hello next couple weeks: Beethoven's birthday, that's right, December, 14, um actually, no I, think of it. It might be December 17th, but 16.

B

Okay, okay, whatever Beethoven's birthday is that's the day. You'll come up with that, their 16th, okay and um we'll decide. You know whether to use that piece of abstraction or not by then the other. Is we have this recurring discussion about augmentation and we have it showing up in both drafts. um Do you just reference an RFC? Do you pull in parts? Do you ask other working groups to augment the line working of the line model and I?

B

Think we've come up with a solution that says, then somebody tell me if we don't have consensus on this solution, but I think the solution we have consensus on is that we will not reference any rfcs for the oam specific stuff. We won't reference the BFD rfcs, but we will ask other working groups, not trill, not BFD, to augment the line model with their stuff.

B

Do I have that agreement right for everybody, except BFD, in trill, for BFD in trill? What will we do.

C

Okay, so my understanding is that will reach to other groups working on OEM frameworks and solutions, to review lime models and consider their relevance.

C

Is that right understanding, because, basically we're not enforcing.

B

The Y model onto them, that's true, if they, if nobody contributes anything, if nobody augments us at all, we are useless with a capital you, okay, I see head nodding. Is that not entirely oh you're you're, nodding, left and right, or right up and down.

J

Jeff has so the problem lime has always had. Is it doesn't really have a good sense of which direction things need to go? You know partially, because we started from two problems. First, one being yang was a very new thing. Second, one being the OEM technologies involved are very distinct things, so I think I'm DFD is anything I know well, BF, DS management, module and yang for the ietf has to be self-contained and self-sufficient for ietf purposes, and it has to also serve in the capacity that it does for all the various protocols.

J

So we have to have something that can actually no satisfy those requirements. Does lime add anything on to that? Not necessarily mm-hmm.

B

J

Lime, making use of technologies add something potentially mm-hmm.

E

J

If you can't figure out what you know that looks like now that we've gone through, I think in a partial, attentional I'm, roughly two and a half different schemes of you know what the relationships of things look like know. What the initial one being by example, sort of know. Why 1731 I triple e flavored? No, it was not something that was a good fit and the authors of no dental that work to try to actually work around that.

J

If you can't figure out what that looks like, how can you ask people to build on top of that.

B

Offers any it's.

J

An old discussion, no that's that's where most of the churn in the documents has gone through is trying to actually find good mappings between these things know if this was not act, I, TF technologies, I think this would be a lot cleaner problem because I, tu and other organizations, but they build technologies, build o, am as a first level thing into each of them and therefore it's clear what they actually don't tie together. Itf technologies. Don't do that now, they're loosely coupled I.

C

Agree with Jeff characterization of that so far in too many occasions we left Oh am hanging as a abandoned child. Something like I can compare only to the mid work that you know being very late always but at the same time, for new encapsulations and that's something that would hope well. I'm may be relevant that work on sfc beer and there are three as that's where we try to come.

B

Up with a comprehensive OEM mechanisms,.

C

In parallel to doing on, networking technology.

J

So, where I think something like lime has challenges is to build that base model. If they'll get this with some other organization, the underlying base models would already be there. Oh and I do you have circuits, and you know they have the only important circuits and all the stuff that builds on top of the circuits. The layers just keep on building up what we do not have an ietf, no spec and sfc as an example. That is effectively going to be a IP technology.

J

There's no IP technology, OEM that we can build on top of so we're. We don't have the same foundation that the other organizations and that I think that right there, I think, is really the thing that line has been fighting with. The entire time is successful. Yeah.

B

Well, I'm is trying to find some commonality among things that aren't really common.

B

Okay, so going back to next steps, um there's the piece of abstraction, that's the easy part. Now how to tackle the augmentation problem, um I thought we had a solution that we'd agreed to that was to get rid of the references to specific oam rfcs and just wait for other working groups to augment the lung data model. Does everybody agree that that's what we agreed to on this meeting?

B

Is anyone wait? Does anyone not agree that that's what we're we agreed deal? Does anyone agree. Remember the agreement differently. Okay, I'll assume I got it right, then so I'll be looking for a new rev of the connection, less draft pretty quickly like soon after Beethoven's birthday.

B

Yes, and with any luck, we can send that to last call in january-february and send it to the iesg by next time and probably the connection oriented one should go soon after that. No okay, do we need a online editing session in January, a interim.

B

Should I take that as a no authors, do we need not an interim in January I.

B

Guess not well. In that case um we have a path forward. I know this has been a difficult meeting. Thank you all for bearing with it.

B

See you all wherever we're meeting next time Tom? Where are we meeting next time? Chicago cool? Okay, see you all there.

B

And who has the blue sheets.

B

Hey, ok, who are you.

K

D

K

Frustrated to my comments.

B

Well, you're talking about the augmentation ones or the involved Lisa.

F

B

Ok I thought that was in the agreement we just rattled off that we would not let the RSI I send them one.

K

Years in before they still they change now, let.

C

Me hit that you make.

K