Internet Engineering Task Force 93, 21 Jul 2015

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: IETF93-LIME-20150721-1740

Description

LIME meeting session at IETF93
2015/07/21 1740

A

Thank you very much.

B

Getting the granted now the official gradient, the line working.

C

C

A

B

We're one minute late.

C

Looks not that that.

B

B

C

You made the slide I'll give it.

A

Thank you very much Ron, so welcome to our line meeting. This is line for the next hour. We only have one hour and we have a little bit of a pad set of presentation, so we wanna get right on it.

A

They note well, which you've seen many times but still contribution to the IDF are not only drafts and emails, but anything that we say on the microphone in the context of a meeting is an IDF contribution which is governed by the IPR loss on the RFC that are cited there.

A

We do have a scribe and jabber scribe blue sheets.

C

A

So, thank you, John a blue sheets passing along and we're ready to get started.

A

Absolutely so, this is really briefly our agenda.

A

And there's something that truncates on the list, so we start with it with it do mini strata via on our agenda. We have we're going to talk first about the young model. Debug is going to be presenting applicability statement. Let you know we have a good plan on on how to approach that slot p.m. and next steps, and before that we want a reminder- and this is something that we want to be very explicit about in words our goals. What are we doing here?

A

What's a scope, wanna review the chart turn in particular the deliverable that we have.

A

If we all remember when we charter this working group line, we have three key deliverables: three main outcomes that were driving towards a young data model which is going to be the first presentation by Deepak an architecture, and that is inbred um purposefully, because we still don't don't- have a lot of clarity in the working group on what that's when I entail and we're going to cover that on the next steps and applicability document which we're going to talk about today as well run.

A

I, don't know if you wanna, add anything to what we're doing here or Benoit.

A

Thank You end reminder also on our milestones, purposefully on red. um The two milestones that we are the winning to deliver. First, the young model and the architecture followed by the applicability we're going to talk. Debug is gonna, walk us through the model itself. The young model we're going to have a discussion on the applicability and on the next steps, go on a cover. What we do with the architecture.

A

So with that, thank you very much and we have our first speaker yeah.

C

Visit you, gender Tommy's, not here, so we want my colleague er the end zone.

D

D

C

D

E

D

Not here not I'm not a presence, a strap, actually we're cutting a gender Tom. There is none of hair, so I ask my colleague cover up the applicability, documented foundation, yeah.

A

The deadly work thing here, yeah, so Deepak I, think you're presenting on the on the young model.

A

Every time you want on a slide that branch, you can just say, click.

C

E

Hello, everybody I think so. I'm gonna cover the line, base young model and means I and chain that to editor, but there has been contribution from a lot of people for to get this model at this age. So can you go to the next line.

E

Okay, so so we had a design team before dallas and we the design team work, was done after that, and this is a status update after the discussion, so mostly more what has changed in the model from Dallas. So so there was commonality between you know. Ipo am mpls, om, 1731, mpls and Trillo em, and this was discussed so mostly we. The thing is: IPO am as a generous in a completely common with all of the technology, so we figure out the common functions.

E

Those are the connectivity, check, connectivity, verification for performance loss and deliver a very measurements where I grieve again. Another thing was agreed upon was that on a management side you can even envision IP as n is in. It has a test point. It is within a domain and and the young extension can be done by technology type, extensions to the model, so so I. So one question that kept on coming in case of IT: there is no configuration of meth, whereas in an l2 model there is an explicit configuration of them.

E

So this is what is clarified that this is more of it. Abstract concept. Where map is believed to be there implicitly for management station to to actually understand that this is the other test point, or this is a sick. This is a sink or this is a source of it test point.

E

So so this we are documenting the line, design, team, finding and also applicability document was to be done so I think that would be covered by Tom's. Like okay, so few comments we got from the mailing list. We are documented almost everything here, so comments that we got word that you know we had to put two editors move, others to the contribution section.

E

There were few open issues. Were there for interface. Referencing will cover in detail in next slide. There is something non key leaf instance referencing to me. This is a ask from us only how we can get it from younger it's more of a language thing. We would want help from the young doctor and- and there were you know, language issues in the graph. What does it means young model actor at management layer that was raised by then, and there is some pee and poo validation error? Is there, especially, it is growing 73 /?

E

So we got comments from author of draft lamb which which Coverdell 2202 l, 20 n, + ug, compile all the comment and send us on the mailing list. We accepted those things clarified on that will cover them as we go along in this side and we have to do an update after the meeting means it once all the comments. Are there there's a special thanks to all the members means who had gone through it, and especially the author of you know, means l0 to attend.

E

Tom Gregg, everybody pitched in to mostly it's more of an IP model. How to actually do that again, open issues where that there, a connection oriented networks? Are there there connectionless network? Is there and you know what is a commonality? And how are we how we address them.

E

Okay, so we had some so what we did in. We had some identity issues that we fixed up up. We updated the acknowledgement, sections security consideration got updated, we had two editors god identified and we generalize the IP ping and traceroute into continuity. Check and part is cover ER TC, so we made the connective check. Rtc optional by using yarn feature, come on based on the suggestion from young doctors and we added support of either proactive and on-demand by adding session type under under rob a small pieces.

E

Again, there is question about the topology and you know to call the independent and the extension of it how to add multicast to the model. So we we are busier. These all things can be added as an augmentation to the model. So we add, we added all the keys so that augmentation can be done.

E

Notification in previous dark was specific to a certain map. We made a general that it can be reported by a different map. It not only two had to be appear: map which will report a notification generalized in abeyance, especially output of RPC calls, did not specific to certain technology type that are you referring to, and I think the prefix some you know, issues warnings and one.

E

We actually split the performance measurement young model to another base kind model draft and that can be augmented by each and every technology in case of IP is a 0, empty, ramp and or in case of trill. They don't delay measurement on this thing.

C

E

Next life, so I think Jen team gonna do our demo in bits and bytes and they're gonna cover this kind of model, I think based on this previous session. We would actually try to identify this OEM young model as a service orchestration model, because it's not it has device specific things, but it is covering a network. So that's why you would need a management system which is which understand the young and it it will build on a metric on client.

E

We also make sure what is the lime scope is because, based on a charter, what we are doing so we said that apology, discovery and you know in capability discovery. This is not part of you know that that tool, or that authorizes the system is already knows about it. What what line is doing? Is it giving who AM data modeling, how to apply configuration for OEM how to generate our PC or notification, handling those things and and again these are things will be covered over netcong and on the device?

E

We would need to have a server and data store, but the OEM entity, which is the existing or implementation. That is not covered by you know means our effort. There is no changes, we are doing in 20 m protocols, anything to I.

E

So again, this is what it's basically saying that you know we will get an RPC call and, on and and on the devices will interact with OEM entity, which will generate the existing. Whether there is a trace, Nautilus, beeping or whatever it is, so we can go next slide, so model structure that we did is that again in sir, we.

E

What we did is that we found a model structure from ethernet and mpls-tp, make sure it can fit in IP world and- and that's the model structure we had done- is more of identification, a test point and are so on generate level. We are saying these are domains. Are there there are maintenance association that the service is there and maintenance endpoints, which are basically can say as test points are there now? This can be augmented based on the technology, and you can put more statistics or counters whatever you want as a as your own technology.

E

Those can be done. What another thing is in our one thing that lime is trying to address is related or in there, so most of the time now they will be able to rest into another room some way em or you know like. If you are going from overlay Network, we have an overlay or em there's an under layer and link or him so model should be able to tie them together.

E

So and that's what the related OEM layer is there on our pc side, we actually talks about reachability verification and again that can be talked about as a thing or what, but we are generalizing, the name based on RC 7276 and base and discussion so I'll cover why there are changes, happen and path, verifications and part discovery and other two are loss and delay from performance. Again notification configuration has been more generalized with a map name and generating map ID to identified and device type an error.

E

All these things are can be augmented by the by the child models or whatever, and we can go next thing.

E

So this is the relationship we are following and so general OEM young model is the base model, and on that you have extension technology, specific extensions. Are there Trillo, M&V, 03, mpls IPR, any foo a.m. and written that also, you can extend it so so again min so it is a hierarchy should be maintained, and you know it can be extended with vendors, specific things also next life, so they were open issue again that this is a funny show from our side.

E

Only that you know we have to representation for leaf interface, and so we choose the representation a and if there is something right or wrong about this thing we are, we can change those things I def will definitely need help from the end of this will open a questions for them that so next to issue.

E

So again, this is again we want to so we have mean, sir. We are trying to reference leave instance. It doesn't have any key or index and like admin status and offer status and even admin status is optional interface lee. So I think it definitely need help from young doctors. I think we got that it's an p on uml plugging tool issue again, we are not an expert, so definitely need to get understanding on this thing and get you to fix.

E

Third, one is okay. This thing happened. A lot of discussion happened whether we are monitoring, connection or monitoring network, and so what we are saying is in connection oriented network. You have a connection, so we can monitor a connection, but in connectionless network there is no concept of connection. So what we are monitoring, so are we monitoring a link or a section or you know, means test point between two nodes.

E

So the conclusion that we are we basically reaching here is that you know means we are monitoring source and sink kind of this thing so between two test point this day and we made the reachability verification as optional, because I be having ecmp you that was also move can be moved. Ecmp is not there in all the way in so there are specific mm pls as a trail as it.

E

So that is not not a function of a common om, so that gets moved to touch and where our IP doesn't have it implicitly in all modes option: okay, next line, ok, so again, miss this question always comes up, you know can be model based on map or lip and it makes sense or not. I think this got finalized. That you know means we can definitely do that and we are treating this thing as a management entity and they are implicitly created and again means now.

E

The question was for the key cmp that I talked in the means: ecmp is not present everywhere so and even mpls-tp doesn't has any cmp, but it has a consult, a map and MIT. So again we are moving this to the from the comment to the.

F

Grandmas k, uncertainty truck, so I just want a point that he simply excluded from mpls-tp architecture, / sake, because it's against the transport paradigm. So in any transport there is no ecmp uh I, don't know how Ethernet OEM supports you simply as far as I know, there is no support for easing theme in internet OEM either.

F

So I think that it's only IPO am.

C

F

That on the IPO am has some instruments that can help with ecmp. But again it's all, in my view, black magic, because there is no guarantee that you can monitor particular ecmp pad and that's where it might be more.

F

Proper would be to look at multi-layer, OEM and then inter-layer OEM coordination, rather than try to do from one layer to look at BCM p. So.

E

So again, that's right in and again what we are saying that drill has a concept of flow entropy and ecmp.

E

We can definitely do it in overlay, but overlay OEM is not even there so again till the time that om, so we could have in future somewhere in doing it. So it's better one can be done it through multi-level to again management system can figure that out. Another thing is that easy and pecan. If I, if a protocol supported, you can put the ECM p in the in the in the protocol yang wallet so.

A

I guess one one way of restating what correct said and Greg correct me if I'm wrong, but I think part of the proposal is Adi. Cmp should not be part of the most generic model and it's you know as needed. It can be included for IP or until age, but not us as for device. Yes,.

C

E

Okay, I think this is the language they should. We had to clarify that you know what does it means beyond data model presented in this chin has replied back to an email, I think we'll try to reference it and give a diagram means. What does it means again means you are we? What we wanted to communicate is that we are not modifying the poem protocol or entity. This is more of a within the young device model and the management orchestration.

E

They have to make this thing work, so in excellent I think we have to run the PN to find other. This is one of the warning, but pointed us from Benoit and I think they have to fix that so next line.

E

So we had given the mailing list. They were other comment that we have accepted. We haven't covered in the slide, but we had said ok to it and we will do in approve of the draft after we get. You know more comment at this meeting and and then you know, means a yak, so.

F

Greg, mostly ericsson, I, just want to point and bring to attention that in if there was there are two documents. I think that related to this work, one of this mpls-tp OEM yang tomorrow and another is ls15 yang, moro, shadow.

E

So, ok, so I think we'll take a comment and we'll look into it and we'll definitely need your help. If, if you know things have to retain some.

C

Clarification question so.

E

Again, it's a so, for example, when you do a thing you know means you are you here, you don't know which ecmp part or you know there is a lot of discussion about you know.

E

Are we covering all the parts in ecmp and to verify that you know and again before him tool doesn't do that we are not trying to do it from the model itself that it can be done from the child, not on the base model, because on the base model, before we used to have a flow entropy where you can put ecmp information and then send it, but that was only being used for trill, so generous OEM model was basically becoming a trill model kind of this thing, so we just have changed that, but but management system can still use multi-level tool and get this thing done so.

C

C

Just pink concept or oil constitutes yeah.

E

Ping is ping is also yeah. Thing is also a mess it for you. If.

C

You talk about, like you, yeah.

E

That's happened: Everett yeah is.

C

A thing and really is the most used. So how do you ignore something.

E

C

E

As a ok, so maybe maybe let.

A

Me try to replace and you correct me if I got a problem, I think what the message is coverage is not guaranteed. Rich ability is part of them all right, so you can taste rich ability. You cannot gurantee cmp coverage on the reachability Texas.

C

Even if you want to go and see it could change at any given point of time, so the connectivity is keeping yes, sir connect. What avail it ok mutant.

C

Which covers so.

E

Yeah, so the basic thing, which is the basis of trace route, so we so those things are there all the base, statistics that that will get even on linux on something those things we have kept. It only thing what they're saying that ecmp with ecmp you can make it more complicated, so we are not putting like management system itself, start changing flow, entropy and start doing. You know those kind of thing before it could have done it, but what's.

A

Relevant area now will run addition to. That is that you know a specific OEM proco that supports of easy and be like an LS peeping. For example, I will expect that on the additional you can do, which ability specifying flow entropy flow, entropy right or entry or in mpls LS beeping, no part of if you do it on IP, you can test which ability, without knowing d, simply coverage but rich abilities. Part of the base model CMB coverage is now.

C

John messenger from ad for optical networking, I'm sure this has come up before I heard a couple of things which made me wonder whether the model represents the lowest common denominator of OEM systems, in which case it's it's and kind of disappointing, but I'm not sure what a strategic direction you've taken in in terms of choosing the features that you get to support in the their independent model. When you comment on that, so V.

E

If you look from the model, we are really definitely covering performance. If you see on the performance of almost all the thing of am to amp or even Ethernet om, all those all those statistics are there even for we had defined features, Pacific RPC, it's there it just that you know means if a protocol we can't change a ping protocol is just somebody just do a normal icmp ping. We can't it be because we are not forced. We can change that thing. So it's more on that, but base models.

E

Also, we will be adding kind of this thing too, so it will get come complete with. You know, base models also, but we had kept that structure to extend added a feature so that you can make it an optional. But if it's there it's an option.

C

And I wanted to comment on the choice to use implicit map replacement and some of the features of Ethernet mpa. Om, I very deliberately use explicit netsertive. You can.

E

Configure explicit map where, as they.

C

Understand why right yeah, because it's a way of configuring where service and points off yes and so it the internet, let we have very specific ideas about where services start and stop and so placing maps of those points rather than and where some other management system thinks they might be. It helps you to establish whether there really is connectivity and the other one that I wanted to mention.

C

Was connectivity check frames and don't only do connectivity verification they do non connectivity check verification so that if you receive CC messages somewhere, where you don't expect us receives them, that's an alarm that you've got a misconfigured service, and so it's very important that that capabilities exposed and the model so.

E

We haven't remove things again mentally add we are influenced by those models. We haven't removed things like what and all those things are there. It made it optional that we are not forcing IP to do what I tonight. Oem has done it so actually.

F

Our Greg Merson again I would like to continue in previous comment that it, in my view, I agree. In my view. It's very important to use proper terminology, because we often call as a connectivity check what is there are continuity, verification and I'll continue to verification, especially a ni p. It's only that, especially in case of S&P. We can only say that we have a route between a and B and we can get there by some needs, and we don't know ah in this particular packet which path we traced when we're talking about connectivity.

F

That's what I think that was a comment about the connectivity check assumes that we can have defined what is Miss connection fact is, and if we look at mpls-tp, it's only one place in IP and I p.m. pls, where there is a definition of what risk connectivity defect is and what is the state of getting in deep black state and getting out of defect? State is so that's why mpls-tp has only really connectivity checked and everything else is a continuity verification Oh. Vice versa, continue to check and connectivity verification.

F

ah Another thing is what I wanted to point is that, in addition to reference to to yank model of being done in mpls working group, I think that might be interesting to look at a little bit more distance, but I think related particular to NPO sdp work being done. We have RFC in out of old secant group and document ready in mpls, ready for working with roscoe and a working group document in the fells they're all related to mpls-tp OEM instantiation, using RSVP LSP theme and extension to LBP, respectively.

F

Okay, so I think that again, since these are methods to do instantiation configuration of mpls-tp OEM and we are using them in yen model, okay, okay, so I think that might be good to keep them in in perspective, because these are already some documents that being worked out and they reflect working groups view on into us. Kikyo am so.

E

One thing is: definitely, we will add a table from 72 76 and with the table will add these young model or the new Ellis, keeping and and look at if there is definitely a commonality or not a commonality, and then rule things.

F

As ours I think that pro, in my view, the common model has to be encompassed in them, not forcing them to change.

C

E

Yet definitely so if there is something you know means if you had to see where that implicit nap can be done there or not, and.

F

Actually again in mpls-tp, because entellus kiki is a foundation for a packet-based transport network, the maps are explicit. Okay, the only differences to compare with the internet, for am ah that it doesn't support levels and there are segments or am supported through.

F

Basically, here are you go OSP, okay,.

E

So, okay, we definitely I think we covered that, but we'll go back again and definitely answer on that and maybe put wording in that say that you know what we covered, what we looked at it and and that this isn't.

B

Long vomito Juniper Networks speaking as an individual contributor, and this might be a question for Gregg Murphy I- think one of the problems that we might be getting into is the word. Continuity has some baggage in our minds, but we think of continuity. We think about continuity among segments that are explicitly wired together kind of like continuity in wiring in your house. Where is- and that makes sense for a connection-oriented assist system. Maybe we don't have continuity in connectionless systems. Maybe we have reach ability.

F

That that is very good discussion, ah but historically we have two terms, and that was explicitly. They were differentiated in RFC from mpls-tp, which was CC cvrd I, so continually check connectivity, verification and remote defect education. So that is where at least we try to differentiate. What continuity check is so continue to check is that you have a way to get from A to B, and if we look at it is so that just normal BF d is the continuity check.

F

But if you provide your identity of your source and you're verifying the source- and you are reporting when you have leaks from other services that was probably mentioned in in the previous in a previous comment right so and you are reporting them, then you are monitoring the connectivity, because you are monitoring not that you are receiving what you're supposed to receive, if you're, monitoring and flagging what, if you are receiving something that you're not supposed to see, and that is Miss connection error, different.

A

I'm going to suggest Greg that you can send probably that a brief summary on that to the least so that we thank you. It's.

D

C

Hello, yeah, okay, actually recounting the comments on the reference. I need to walk. Actually I should I think regarding the 4x comments on the Emmy airport. I think we could look at that and we we sink any fo for more recent at a specific, so wait. What wait? The proposed hey actually will try to do a complete depending the model so that that's a different. We love the little behind.

D

Today said model to say: if they're there something we can kick.

C

Would you regard this is all I can get as close as the one? Do you hear me? No, no any collapse.

C

Okay, we do Cucaracha telecom I'm new to this I rushed through some drafts and follow G presentation. My impression is that Oh am in the sense you work on its data, plane, verification, where's,.

D

C

Om for an IP network would be also to to watch condors of interfaces and stuff like that nice. If, if you, if this is out of scope, that the Charter would say that, but you focus on data plane, verification thanks.

E

So I agree, one of the things from operated is definitely looking at the counters of the interfaces. But at this time we had focused on om, but we, if we.

B

C

C

Agent Ferrell I just wanted to when ron was talking about sort of continuity in connection this world.

C

It's not just reach ability, it's lossless reach ability, because that's how you get the equivalent of signal degrade or or broken connectivity.

C

A

Very brief, because we will.

F

Email, the other exams, harrison AR. I think that it. Yes, I agree to the include active and passive om.

E

So, okay, so I think we would want it to be adopted. We can work on it as a working group document. I would want working group to look into it, see it's ready to get up, did it again, model can be extended and we can call at this time is doing data, plane, verification and if, if it is required to look for us for counters and that we can actually look into that.

A

Who has not read the document.

C

And not not focus.

A

Or not ready, who has not ready? Who did read it so so so the majority of the people in the room, algae have read the document and there there's been a load of discussion of it on the list. A lot of iterations on revisions. I, don't know: should we ask for.

A

So so lets us this way from all of us in the room who supports adoption of the document. I say our line working group document.

D

Yet is speaking on the air talking about 205 buzzer, which is a 6 button, which one.

A

I'm talking about the document that we just presented, epoch just presented on the young model, not the applicability, not not what is showing right now. The teaser draftees are blah blah without the performance management.

D

And I are you speaking and I made a notable comment, and so maybe they oughta set and updated. There are six.

A

And the baby, I think the easing of the we should think.

D

Of on the dev 6 button for adaptation, there five leads to the number of issues and I do not find any fixed issues.

B

That that kind of makes sense. Why don't you post the version six you talked about and then we'll have a call for adoption on the mailing list. So yeah.

E

That's fine that makes sense.

C

G

Yeah, hello, everyone and my name is yen Jang and the list on the first actually actually before the tub, we're presenting this presentation, but unfortunately he's here. She is not so good, so he can not make it so I were instead of making this presentation, it's some work about the applicability we can. We start with some background. This work item is in the Charter and which means the group has to deliver some applicability document to prove that the usability and the extensibility of the line based model can be used for the specific OEM technologies.

G

So since the base model is kind of stable so last time the Chiricahua volunteers for this document, but we noticed there is no output, so we our colleges in our co-authors, work on this and provided the roster version of the applet II. This is the moral point of view, so will welcome the comments and the questions on these documents. So, let's start with how we organize I, how we organize this document we are. This document is mainly talking about it. Two parts, the groceries.

G

We provide some guidelines to extend the base model to generate a muddle, those specific OEM technologies. This is part we introduce.

G

We use the identity to define the new OEM technology types, such as only our trio OEM, which is not to defining the base model, and we also use choice case, a statement, sir, which means we add a new cases for specific aureum technologies to bring their functions and parameters and the easiest part I would like to say how to extend the base model is first, you have to choose the right layer, which means you have to choose mb and they MVP I wish they are. You want to argument and there's.

G

The second thing is you have to choose the right angle point and the last one is defined and inserts the date models are dead, notes for the specific OEM technology parameters or functions, and the second part is we resist kind of guy denies me describing the first parts. We demonstrate the usability and the extensibility of the line based models with specific examples. We take some examples, such as IPO am 300 m and n bo3 or ems. The TTL things of these OEM models and the examples can refer to our chops. If you have time.

G

Thank you next slide, and this is some questions we have this group so most important. Why is there rice go? Are we know there are many OEM technologies in ietf, so we want to ask the group: should we provide demonstrate the order? Are we having technologies are to the demonstrated the usability and the extensibility of the lifespan by line based models for all this specific Orion technology, or we just pick part of these Orion technologies to prove the line based model can be used and extend.

G

So the that's the most important thing we want to ask the group what kind of example should we take and what things we should describe and defying this introduced in this document. So the following are some open issues which actually has been discussed in the parks presentation. The first one is connected and their connection is, and with the signal rise, some optional features, which is not too defined in the generika OEM model, but maybe defining some a performance measurement medals or as some optional things.

G

So we wanted to see whether we have to demonstrate this optional features in the applicability document, and there are also some items are described, seeing the draft as the open issue, and it should be, this discussed Thank You Nestor and the next step is. We would like to ask the group what we should do for this document.

G

Yeah. I think someone is already standing it there. Yeah thank.

F

You Greg misty since I was mentioned, so I think that, yes, it would be very good and if use cases will be provided, the demonstrate applicability of proposed model to other technologies that ITF originated and that would include IP, IP, mpls and mpls-tp and in particular, since I already mentioned the work on mpls-tp, I think that will be probably easier to do.

F

There is no work, as far as I know, on IP OEM, yank, moto and 4i p.m. pls offers of mpls-tp, or am young model intended to start this work by extending mpls-tp document. So we can work on it together. Yeah.

G

Thank you very much. Actually, as everyone saying, we would like to say this examples are not some standard specific OEM for the other technologies is just there imposed you to teach people how to use these base models. So specific OEM models will be dying their respective groups. We just provide some guidance and how people can use it. So welcome people to help us to finish this document and make it a complete one.

G

So do you have any questions? I.

A

Do have one yeah, hello, I, don't have one quick question. What this document is trying to do is prove a proof is too strong. Maybe what demonstrate that the base model can be used in a number of different technologies? Have you connected with the authors of different draft in different working groups like the BF dls beeping mpls, to get bear views on whether disciplic ability applied to the world that they're doing, because it's ultimately, but that would be the ultimate goal? Okay,.

G

So ah yes, yes, we should contact the descender specific om model Sorcerer's to get your feedback on this exam post by yeah. We will do that where try to do that.

D

Actually, we pretend to do to connect the order: okay, yeah feedback, your question or two: we plan to container also the relevant om young datamodel chapter after we can get a nine-piece model adopted its well convinced the bottom. We can work together, otherwise I think we may how'd you convince these. Give us two yeah.

F

So I probably can represent at least three design teams, BFD and OSTP OEM and LS beeping, and I think that the sooner we start discussion that the better, because these documents are maturing and probably our station would be that their common data model will encompass sing and not enforcing anything on the models that we're designing.

G

Thank you, I will work with you all these documents for further discussion. So any comments.

A

Thank you very much. Thank.

G

You thank you very much.

C

Have another presentation.

A

We are low on time see if you wanna make a point about.

C

The architecture I'd rather make it now I.

D

Think, okay, in the same pronese same slides, send my model I'm pace model John.

C

D

C

D

Youtube you can chat on not this one.

D

9-Piece model: no, this one.

D

Yeah you scroll down yeah.

D

Okay from here yeah yeah, yeah I'm, going to talk about performance management model is actually the companion companion chapter ready to the line based model chapter actually base a discussion, a razor compatibly we are at the beginning. With talking in this chart, we sink a proposed amendment may be a small portion of dates in line based model by the audible ratings. Obviously 7276 we found another all the OEM technology support performance. Amendments alway prefer to actually split the votes Melinda from the line based model in that and the document in a separate chapter.

D

So what is the ruins the moment about? Actually we? Actually, we can have a management system to a performance woman from the violence amendment to buy issues, a kilometer to the nettle element and triggers that nettle element to excluded the columbiana get is a prunes, many results it back to the mainland systems.

D

So here we covered two common pieces when sell as men. Another is today in a Amanda. It was a week when we we found the when we the boom technologists bother. Most amendment is two men in it, and I need to be a meditative to supporters, so we're using the metastructure proposed paying the land based model which, as defined in the writers and component to insert it, that they did not a leather, relate to the bones name and and they're. Also.

D

In addition, we defines a a pc, so the last name in the film a man. So here is the overview next well. So so what is the component? Which shows? Actually we extended the land based model we weeks and met with a configuration with the last name and the lame and the photo was named in a way to find the role of the testing point and the name and.

E

D

Soda, so for the road test point, for example, we using the name and the tablet to distinguishing when we deliver one women from the two-way moment. We also using the session copper to distinguish you the county, bound amendment from the proactive that moment and in addition, we actually defines a name and a method. We cover the men in the interval started time and any time and also as a parameter. Yeah further delay is similar.

D

The emitter can be applied to the delay Raymond next for the RTC extension we actually in choose introduced at for a pc to for the loss meant to for the delay may endure, for loss may mean that we reduce creator last name and an approach to Amendment and the same for the delay. Man mandir. So next.

D

This is some open issue. We discovered when we talk amenities, chapter and a way which is cut with some calls or like Tom, and we he and actually district this country more related, inconsistent usual when we, when we write the young model Dakota way.

D

We, since we just assume that the porter lost men and I bought a delivery and only support of the only moment, but this is not true, so we need to fix this inconsistent use since this loss, men or killing initiative supported posts on the mountain and the up grow a cube, so we need to fix this next.

D

The second issue is about whether the province amendment always a part of the line based model. Actually, this are a little bit sugar and the I think we have some argument here. We think if the OEM yamoto support the performance, a man that can be using to do the end to end a pass characters gixxer to connect these characters, the most name and it a man in a mask- is about also for the purple team, and actually this is kind of generic model.

D

So we we sing it's better to have a single Java to put it as the model in a single eNOS in the same place in a single document. So this this is just our opinion. We wonder a solicitor working opinion on this.

F

Gregory Erickson I will suggest to look at the work being done in the L map working group, because its reference model includes controller measurement agent in a data collector, and it's already has informational model in the data model in a working group is working on the protocols to instantiate measurement task as a combination of reference to measurement method and parameters, and it includes the schedule which defines their career de ce and.

F

Period that measurement task has to be performed, so I think that there are a lot of commonality and we should try not to duplicate the work being done already in elmet group. So.

D

Thank you for a pointer is our network. We haven't rated that Java, where we all take a look at that check out to not.

B

D

B

By mercker speaking, is individual contributor. I'd argue with the must in your pro um could is a possibility, but it would also be a possibility to leave place for an extension and put it in later. That way we could complete our charter and move on and maybe recharter when we want to do this later. Another possibility is, we may need to go to other folks for expertise on what what the performance management metrics should be. No, maybe it's something: maybe it's a box that we want I ppm to color in for us some other group.

C

But much less so I want to combine it to commence from Greg and from wrong, because it's I've got to send feedback. What you're trying to do is the RPC there is we do what I ppm has been doing and they've been doing that for as long as I can remember right so, and the second point is that maybe the things that we do for punishment aren't just Technic, but there is not much for something generic so having just something a placeholder. If we even need that I believe is good enough.

C

So I'm not too sure about this. This work, okay,.

B

Thankfully, what to do what to do next.

A

B

We should probably talk a bit about next steps. The first next step for the yang model arm deepak you'll push our post version 6 fairly soon, we'll all take a look at it and review on the list. Maybe have a working group actually definitely have a working group call for adoption. Second issue is the applicability statement that seems to need much more work. So would it help to bring the authors together between now in the next IETF? For you know some kind of brainstorming, good idea.

B

Okay, we'll set something up arm for the authors to me figure out what is that they meant to sent them. Do some massive work on that document. The next is the architecture document which we haven't even begun to think about yet, and we don't have a volunteer to to work on it, yet um it would be a good idea to start a thread on what should be in it. I'll kick off that thread on the mailing list. We're going to be looking for some volunteers to author that um what do you know?

B

We have hands going up and I'll work with with it. Whatever volunteers are their own, given that I think we've wrapped up our agenda, but when I am.

C

But Mara speaking so us mentioned my thing by Carlos. The end goal is that would have this base young model and the specific extensions and different technologies right, and there are some of them as Greg mentioned, so we should make sure they did this work together right and, as a consequence, all this you will get your architecture document that created and yep. You can BTW and everything.

B

And given that, if nobody has anything else, I think we've wrapped up our agenda, it's Miller time or hang on we're in the Czech Republic. It's time for goodnight beer.

A

Thank you very much. Oh and the other blue sheet.

C

C

B

C

Oh I, sorry, thank you. So much they'd rather remember what stuff now.

F