Internet Engineering Task Force 100, 14 Nov 2017

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: IETF100-CCAMP-20171114-1550

Description

CCAMP meeting session at IETF100
2017/11/14 1550

https://datatracker.ietf.org/meeting/100/proceedings/

A

B

No animations were you.

C

Okay, guys, it's time to start yeah.

C

Okay, good afternoon, everyone welcome to circumcision. So actually you know arm this time. Agenda is really very busy and Deborah asked me: ask her to cheers it's bad for us to end the session ten minutes earlier and then for people to take the pass. You know to be more time for for the social event, so yeah I will quickly go through with some general information. That has been your not a pity them 100 times, maybe: okay, okay, load, where I think you know it were very escape.

C

So uh please big in front of the mic and state your name before speaking, and also for the minute taker. It will be much appreciated if anyone can contribute the minutes over there. You know our salute the use. A pad, yeah, okay, III, think how men can have a list session this one. So you know actually out this time. We have plenty of jobs, actually twenties range arts 15 slots, so the agenda is really busy and actually that very good that they came to steal. You know glowing.

C

So for the IPR process, yeah, so the contributors- all you know closest you have to you- know the price IPR call as soon as possible. Otherwise it will, you know slow down there. You know table G post and we always encourage people to use the Mentalist as as much as possible, especially, for example, to introduce some new topic to discuss some technologies and also especially, we need to use the many lists to build. You know: table chicken, sensors, okay,.

B

Okay short update on the status of the Working Group. We have no new Avada, no new RF Caesar since last meeting. This doesn't mean that we have been lazy because we have two drafts ready for working group. La Scala, APR polling is already done and we have a one draft ready for working group adoption polling.

B

We have one dropped in the editor queue and wander one in the ASG processing that piece okay. So we have five drafts on the agenda. First, one is the microwave framework. This is one of the two documents that will be lost called immediately after the the end of meeting, and we will have all the young models for w.zahn, microwave and OTN presented. Today, then we have the framework for WDM interface, management and controller, which is the second draft that will be last called after the meeting the next third. After that we will progress.

B

Is there rsvp-te bandwidth availability? We will launch a PR declaration request after after the meeting we have the transport and bi use cases which will not be presented about. We have a to use cases presented, presented that they, the w.zahn impairment, validation, information model which doesn't have major changes compared to the last meeting and they OSPF available extension, which is in the isg processing no liaisons, neither incoming or outgoing, which is a quite stranger. And that's it. Let's start immediately and try to save as much time as we can.

B

Unfortunately, the this is a wonderful thing. It doesn't work, so you need to ask Fattah to move the slides for you.

D

Hello, everyone I'm going to present the to working group dropped its those dope ronzo microwave presenting the first one is on framework dropped, so the changes front. There are one version.

D

We add some tears to indicate that in the young job, so we decide to move, some parts could be generated for other technology use and we also add some tests to describe that way. What we have done in Astra, IDF, hexam and another appetit ism. We have a new co-author, the near now from Nokia and immediately when we finished update down there one motion we found that we left empty on the Security section, so we have another 0 2 version. We fill up security session and also fix some editorial needs.

D

So there's no technical issue opening for today for this rim of dropped, then non-thinking. The nested will be a working group, not code, ok, so the second one is under young module. Ok, let's see there are two main changes from the other one. The first one is that in a three-team will receive some comments to review the model.

D

If some part could be generic for other technology use, then we revisit the model and we decide to separate the interface protection parts and the generic part, so this one can could be used or by other technology if they're interested turns the interface protections and then the list of changes that we also update the models to make it the MMD and compliant.

D

This slide shows the details about how we use that's the protection part, the left side, the generic interface protection, and it's defines some example, protection types and the working entities or some other related or protection related objects and in the microphone radio link part we make use. Actually it should be. The arrow should be right to that direction. We make use of the interface protection and module and lets you use that in our microwave specific model. So this is for the protection groups, another updated and be a companion.

D

So it's just reorganized and there are rearranged.

D

That's it for the update and the author explains that we have finished all the technical work in the model is being complete and stable. So we would like ok.

B

Thanks a lot for being so pickup, so before moving it to questions or comments, we had we had the chapter with with Deborah with the members of the design team, and we decided that closer the design team. They didn't excellent job with also a prize at the ITF. Our Catan was.

D

B

In Prague in Chicago, Prague.

D

B

The deliverables have been delivered, they've been adopted by the working group and they are.

E

B

And closed toward the working group, La Scala, so thank you very much or all they are the work that you did now. This doesn't mean that we are not accepting any more work or contribution on microwave. Obviously this was just the starting point and that's it yeah.

D

I actually also, we are also considering for next step to visit the topology argument for microwave. Ok. Thank you. That's.

F

Okay good afternoon, my name is Joey. I'm from Maui will, firstly, introduce the changes of odeon apology, graphs in the OT intifada draft they are as varied as in the introduction part, we had a clarification about the interface, the interface independent and the applicability of the models described is described in another craft and for the young model, namely attributes are removed and as they are covered by already by the TE quality model, and the client facing related attribute in the LTP are also removed, and there are three attributes: TPN range. Yes,.

G

F

Ts tree and added it to the link attribute.

F

In for the newly added attributes- and they are especially for the TPN range and the TS range I used there for the inter domain scenarios in the inter domain scenarios, for example in the a CDN Seng in the context of a city and MBS, a need to configure the inter domain beings. So it's the ambience II need to know the information for the for the inter domain names and the each PNC.

F

We all report informations like the available TP and available TS range and similar for the ps3, the granularity for the link and- and now this here this is the yon tree.

F

Currently, there are still some attributes lottery to be removed as they are covered by the teeth quality model by the unreserved bandwidth attribute next step we will removed and then currently this model is an da compliant and so for the next step for the OTN topology model, we will remove the attribute, which is a complete outcome covered by the sheet polish model and to align with that heat water model. And we are thinking about whether to report as a multiplexing hierarchy of the inter.

G

F

As they are, maybe a different multiplexing hierarchy is on different domains, or we have also justice simply assuming that assume that they we only consider the single stage. Multiplexing is also okay, so it.

F

Is open to open thought or the working group you can discuss about this attribute and the next phase and for the Odeon panel model the OT internal model?

F

The changes are simple: we, along with the current t panel model, which is an MDA compliant, and we add an RPC for fast computing and they renamed the I transport types module to the Odeon types module and, and that information in the in in the module is a key part. Changed here is a young tree did have tree for the ODN tunnel. It is compliant with MDA.

F

Next, thanks and for the newly added RPC, it's a used for pass computing and the input part. There are three parts which is a general input and constraints for pass constraints and label constraints, and some of some of the attributes covered in the covered by the past computation graphs. So the next step will align with the past computation graphs to see how to to revise or just remove, the sum of the part to the past computation drops and next thanks and for the input part of the RFC over RPC.

F

Sorry for the input part of the RPC is here as January input part such as a request, ID source destination and priorities. Such kind of attributes are already covered by the pass computation draft, so I may remove this attribute and for the past country constraint, part the in the for the primary pass or the secondary pass. We used the ER. All explicit a-rod's objects are used to describe the second pass.

F

For the label, constraints is same as OT internal model. The data model to to describe the end of the tunnel or of this path or the end of this path and for the output part, is also represented by the explicit at all. It's pretty round objects for the poster proposed for all the primary pass and the secondary pass, and so the next step we just we are trying to align with that. Has a nice welcome. Any comments.

H

F

Okay, so it's for the audience large model and the low T internal model.

C

Just just a one-click comment, I think: besides you need to in live with the past commutation chapter. You also needed to alive it. So you know our Kitano chaffed.

I

Hi, my name is ian. I update w, send young model next yeah I think because of an MDA issue is finally, we are NMDA, compliant and, and actually we also introduced a TW sent tight model in this draft, which is another young module to place or W.

I

Send specific information such as W sand, node types and the application code, for you know, based on ITT on G, does six nine eight two four, the comparability for the proprietary interface and wavelength assignment policies or under TW n type module, and also we cleaned up so that we only argument w10 specific link attribute from theater polish model. Yes,.

I

Yeah, so this is a basically w.zahn topology module. As you see, a lot of augmentation from t tunnel model achieve a topology model, and only we just kept. What is a unique for w son next, please so I think most basic work is done, but possibly impairment data or we can integrate introduced as a link model in the next revision.

I

So that the mdac can use those impairment, data for past computations, so this is one thing that remains to be done: okay, so it's not quite stable. Yet, okay, all right any question on the policy model.

J

Dieter bell or a Nokia I have a few comments. There is no definition in the draft regarding the channel numbers. I think you are defining those in the wavelength range type. So what are the channel numbers that are they used in the draft, how they are defined so.

I

Which one you're talking about.

I

Channel number.

J

W.Zahn yang's right so.

I

Here, a young mother here, right yeah, which attribute are talking about.

J

The channel number channel.

I

Number, yes, you think: that's in 38 apology model, no.

J

I'm asking where is the defined or.

I

What is channel number okay? um Could you flip, through a young model, for that I? Actually, I forgot I need to look at my young mother.

I

Yeah but we can talk offline, okay,.

J

Yeah and I also have a few questions regarding the grouping serve.

A

J

Some link, attributes I, think there are also some issues that should be discussed. We can take that offline I, don't know whether you're okay.

C

Okay, so yeah I have one question. One comment so look at in the next step step. You said that you are going to and some impairment data on, topology young, you know. Actually this kind of information may depends on the encoding or the impairment that you know the encoding for the impairment. Actually, it is individual jobs. Mayer sometime, you know to move forward the encoding jobs. So if you are going to add this kind of impairment, information to their rubbish on topology, maybe also it take it- needs to take much longer time.

C

I

So we can either move without this and then later we think about augment right.

A

I

May be better consistent with gentleness approach like.

B

I

Okay, no problem, that's even better.

I

Yeah, this is some tea over this internal model.

I

Please yeah, so we did clean up quite a bit. Basically, we argument for things. We argument teeth tunnel model for basic configuration and global pests, constraints for W and wavelength assignment policy, and we also augmented ITF OTN tights, because oh ten types now defines client signal for source and destination.

I

So we think that that's a good idea to augment OTN type and another one is um IETF TW, some type for double wavelength of shannon policy for configuration, and we argument IETF t patch computation, RPC mode for stainless from each other and the sergers documents, so it compares it medically. So that's so happy with that, and then this is NMDA compliant model next week yeah. So this is a basically you know: I TFT W send you know. Tree looks like, as you see for organization.

I

We only specify what is specific to W sin T next, please. So we believe that this is a good base for working group. Adoption, along with 30,000 topology model,.

B

B

Everyone wants to go to the so shot. Okay,.

I

Okay, thanks, you will go to the list. Yes, okay, now.

B

Actually, yes, two words about that: we we want now to speed things up on young models. Okay, this means w son flex agreed that Oh tiene microwave. This is this is day is priority because the market window is open now and we needed to re-up.

B

It's going to save us the 10 minutes with the speed.

A

K

Everybody a little woozy from away and providing some status update of the transport MV I, designed him like this okay. So this last summarize the goal and deliverables as patty from the design team and our working methods. I think we can skip it next, one.

K

So what we have done, some work on the use cases that we are analyzing and we have a treat basically drafter one rough that describe what are the promised scenarios that we want to honor, for which we want to see how existing your model can be used, and we have a two drafts which are describing in detail how these models can be used to address specific deployments, in particular single layer, single domain deployments and a single layer multi domain deployment.

K

The first draft is about the generic applicability statement, use cases which has been recently adopted as a working group document, and we have made a few minor change to clarify the intention of this rafter.

K

We don't see major open issues at this moment in time on this document, and we have some as a next step of what we are looking for is to do some cleanup of the text of all the enemies, understanding of what we are talking about, and we are looking for further feedbacks from the working group next and for the use case number one so, and the analysis of single domain, a single layer.

K

We have published a revision from the last ITF meeting, where we have started the to analyze, how the model can be used to support Ethernet services of interim point-to-point services, IPP ln, v, PL and OT n client. Like a statement at the channel over at Jana, we found a some open issues on how to use this model, and we have some questions about how the T topology mode attributes and the ITRs topology attribute fits together. We are going to analyze.

K

We have a minor addition about how to integrate edges on code that we are developing into the rough, though it's going to be fixed in few days. So the next steps is to resolve this issue and to complete the document and for them for the first issue. I ever dedicate his life next one for the use case tree, which is a multi domain, a single layer. We have a new draft submitted in this meeting.

K

We have started to analyze. One of the major problem with multi domain is how you stitch together the inter domain link. So the inter domain link is reported as extended domain links by two PNC's and the MDS. He has to be able to understand how they are glue together. We have started to good discussion, but we have not finished so we have to finish it, and then we have to complete the rest of the document. Then next.

J

K

Sometimes, on the open issue, so the big one has been on the multi domain is an inter normally stitching. We have been able to thanks to Igor as well, and the apology so should thank analyzed the plug ID, so the Plaga D has been analyzed in two particular and that we see that he supports two cases, one case where this number is assigned by a central authority and configured on both BNC domains and reported NPI to the MDS C.

K

The second use case is when the two H devices Excel to discover the link VL NP and the information that is out of this cover is reported to the mdac and by by business-casual. We notice that the Plaga decent asset to be changed and that the topology has been changed based on our. So we have delivered a feedback to the tea topology model. That's one of was one of our tasks and this based on this feedback. We have this. The structure is allows also the coexistence of different mechanism.

K

Also, you can have a central authority on something's discovering on the other links. There is no conflict or you can have also different outer discovery mechanisms running in your network, and it should be everything should work with plug ID. A second issue in a second option, which were we have not finished, we didn't have time to analyze before this meeting, is that the wheel can associate a configure either on the PNC or on the BSC. How did he know that the LTP ID on one side is at match to do that?

K

You know that the LTP ID on the other side, the configuration on the BSC, is already the describing a topology I did from tipster, and there are some issues described are the other case is not described because it's more an MDS internal configuration. What we are asking now we are to analyze this one and we have to understand whether there are some issues with the option of black ID. That is analyzed.

K

That requires to look for other mechanism and we need to understand what are the pros and cons of this alternative solution and in case we have more option what happens in case for interoperability. Different pnc and MDS C make different limitations of what is going to happen. So these are a major issue. I have.

B

A one question from there on the plug ID: does it allow to understand whether the relationship between the two topologies is appearing or client/server and.

K

Well, I think the link is to be paired by definition, the inter domain link is a is the same layer.

K

Okay, so any input any comment on this open issue from the working group.

K

Okay, next one is about the client services, okay, we, we have looked at the scenario and we have found that with these working documents, we have some questions that is not easy to solve, with implementing quality time. Grantee topology. The first question is on which topology, Datanet or other of the end clients, academia and fibre channel assessing, should be reported in the ocean topology or should we have other topologies?

K

And if the f Jesuses links and we have the audio tunnel, we had to configure the relationship, because the t tunnel is between peas and then we need to know which SS link goes into once you set up the t-ten. You have to say on this. This service is an association between the SS link and DT tunnel and also in case of EDP L. You have been an ID classification criteria that has to be clarified, configure be the number one goes over tunnel number one below number ten goes over 10 and under trend.

K

This information is not in the in the current e Mullis and we have analyzed them. We have started to analyze that two options. One option is a set of draft that will be submitted to this working group, which will around the agenda, and basically those rafters mean triggered by the discussion, and another option we can analyze is the. What open config is doing. That was raised a comment that, let's look at that and we have not yet started the complete an idea. So we are not taking any decision on as a design team.

K

We are open to suggestion or comments from the.

C

Any comments so I have one comment, actually I think actually on I would like to suggest the author's on design team to take egos jobs at input. You know, because that job also talks about use cases and describe how to use the young models for the transport network, so that's good, a foundation for for the for the Dan. Thank him to take a Tim 30. Yes,.

K

Yes, we we are taking it, but we will take it related to the references we will see what there are, whether it's missing there or whether absolutely yeah okay. Thank you. Okay. Thank you.

H

Okay, hello, my name is Tommy and I'm going to present her. The three works about the young models for odine client signals the respectively, the service model for protein client, the topology model and the tunnel models next, so, firstly, we are going to introduce about a service model and a service model which is different from the previously defined the topology model and the tunnel model. So the most important issue for here is: we need to understand: where do the different models seat?

H

So considering a multi-layer network scenarios, there is surely a higher layer as a client and the bottom layer as a server. So usually we have defined separately for each layer. Where is the topology and respondent on?

H

Oh by doing the tano is newly configured between the the boundary of the the nose calling the TDP, and this is used to configure the network considerably the kind of service it is surely defined the in between the URI reference point in the network, which can be the kind of reference point between the client node and the summer note for the inter layer link.

H

So this means that the the server tunnel is early used to carry the client service and, as the tunnel is, has this kind of relationship, so that I know is different from the finest service and they should be described in separate models.

H

So here we are using an example to describe how it is applied between the different controllers and we use the ethernet over OTA as a example. So considering the ntsc and the PNC interaction, the first step will kick off the request of Ethernet service, so that is the the client layer stuff and correspondingly, given this kind of route past, the pmc would respond a messenger to set up the OTN tunnel.

H

This is how the server layer would respond, the staff and given this tunnel setup, so there would be a kind of virtual link, which is the dashed line between the internet service, that that is a part of the client layer and appears dependent on the OTN tunnel, and there would be corresponding topology changer, which means that the controllers also need to update, as there is an anthropology, and if we only focus on the Ethernet perspective.

H

We can see that, given this kind of appearance off, isn't a link we need together to update both is an anthropology and then we can trigger the tunnel establishment based on the Ethernet link, and this kind of tunnel can be used to carry the service from the upper layer of the Internet.

H

So currently we use this table to summarize how many potential service tab can be in the in the client service. This is comes from the RFC seven one, three, nine about the otm GP and the currently. The drafter only contains the Ethernet details and we are going to develop other kind of service types in the in the future version.

H

So then, we'll we would like to have a very quick, real after client, the service model and our young tree. So the first one is the client service model. So it is mainly used to requesting the client service by specifying different service attributes like very global, high-level descriptions and the service type.

H

And then this young tree is shows the parameters contained in the internet quality model, so this can be used as use the win-win to athletes as a client that layer topology- and this is an augmentation to the teeth of polarity model. So in this work we in this current version of the graph that we use the Ethernet header example came.

H

So finally, the model is about the Ethernet hano and this model would be needed. The ones there is tano in the client layer, so again Ethernet in this draft.

H

So there are also a few open issues in the current draft. The problem is the first problem is currently only the Ethernet service is included, and we need to add others and all also for the Ethernet. So we also need to align with some other young models in the ITRs model and then mode will work, because they also have some kind of layer to stuff and together with actually 802 dot. What they're into we're not sure if there are different client models like SDH.

H

Ethernet and kind of fiber class, whether they have a common partner to be generic or there can be separate models. We will firstly define those kind different kind of models and find whether it is and currently the the model is still not and the a compliance, but we first want to clarify the technical stuff and then make it an MDA.

H

So here we would like to confirm with the working groups as the work is useful, and this is actually as it hello, president theater. This is triggered by the discussion of the design team and though we would like to work together and fit into the use case in the scenarios. That's the perspective. Thank you.

D

Okay, I have a comments, so definitely I think this is Amy from Huawei. Sorry, so definitely I think this is a very useful work, especially on the Annette, because also isn't it it's a very one of the popular client service, even a microwave link. So we would like to say such work and I also have another comments.

D

If the one of the common use case in our case is that we would deploy cooing Cleo, isn't it Mordo Ethernet service, so I'm not sure if the current motor have addressed that or it should be addressing in future.

H

Currently, there is no come cool stuff so that we can integrate in the filter. I agree with you that the microwave is a potential server to the Ethernet service. Right, yes,.

B

And, and not only microwave I mean I I see these I probably already do these comment them and on the mailing list, probably generalizing. This worker is something that is really really useful. As usual. My suggestion is to try to see what is the generalized part of a bit to bring it to this and then do all the technology specific extensions in it see camp you.

H

Know yeah I agree with this approach, but I would like to emphasize that this work are basically augmenting the te generic model, so it's the the scope would be still 40.

I

Yeah I think this is a good start, but in regards to you know layer one. Actually, we have a layer, one connectivity service model so.

B

Dissipated. My second question right.

I

So I think you know you don't have to do everything here, but focus on internet layer, zero service if it comes along I, think that should be separate service model. So I think if you focus on internet, that's good base, I think just to move on.

H

Page two yeah I agree actually in the slides yeah. This page will focus on the network side, and this is our kind of supporting the Interop on the NPI. So I don't think it's a conflict with those kind of Ericsson service model, but yeah yeah, but I would like to I tend to agree this, because there is a slightly terminology issue, because both of them seems to be called as service model, but actually I'm, not a expert for terminology.

H

So we would like to hear from the working group for whether there is some better name or such kind of work.

K

Boots from away to comments, I think for Q&Q, if it is a queuing, cute point-to-point, T tunnel, it's the Internet tunneling that is supported here. If it is a connectionless Q&Q, then we don't have yet analyzed that that that option so depends on which type of Q&Q implementation you are looking at and I support. I mean this one is for the MPI is not for the CML violator. One CSM is for this, so you need to translate the layer on CSM requested at MDC into these commands.

K

That has to go to the PNC to set up the physical network element, and maybe we need to align us something in when we go to the generic client to make sure that the same idea is used to some extent, so the insane client is called in the same way. Otherwise you have too many translations.

L

Document just ensure number one I Triple, E 8, or that one is working on Ethernet and Q&Q and anything relevant to the workshop, probably be aligned, I'm, not sure what they're and the work probably should use the young cattle of the check, but it should be aligned number two I tourists we know is going south, so you had the point of aligning with a tourist. If I Taurus, it's not there and drafts are in limbo. Maybe we shouldn't we need to figure away. So it's not a liability for us.

M

Good morning, good afternoon, this is a presentation about the young data model for layer, 1 CSM, so layer, 1 VPN. The starting point of all this work was the RFC 48 47 that provides a framework and the service level requirement and injera describe all the deployment scenario for layer, 1 VPN as a service. So- and this draft want to make a young data model for a layer, 1 VPN.

M

We prefer to call connectivity service model because we cannot talk about layer, 1, VPN as a service, so in order order to make the difference between layer, 1 CSM with the layer 1 to SM and layer, 1 and then live to a layer, 3 sm. So next this is the first deployment Shannara. That is a simple.

M

This is very simple, because we are talking about a customer that won't layer, one VPN services to a provider, so the layer, 1 CSM in this case is a modeling of the interface between the customer service, Orchestrator and the network controller. So this use case could be not so common, but is useful to explain.

M

Also, the second scenario, if you can go to the next slide, where we have a very common use case, in particular in the same organization, we have a service sense, Sdn, orchestrate or services, the uncontrolled for layer, 2, layer, 3 services, and we have the network as the uncontrolled for layer 1 at the end. So in general, this is the use case where we have an layer, 1 VPN. So a common.

G

M

A common transport network layer that is shared by different services and, in particular different services, can be different service department, for example, of say, organization or same service provider and so on. So this is interface, make sense in this in this case and make also the difference with the layer, the layer to SM and layer 3 SM. So if we go to the next slide, we can see these difference.

M

Ok, the architecture is a monolithic architecture, so it's very simple for for now is otoscope the multi domain and multi service provider, whose case, but just to make you to understand that we have the layer 2 SM under left years, and that is the interface between customer and services. The uncle Orchestrator, while layer, 1 CSM, is the southbound interface between the Sdn controller and the network controller for layer, 1 VPN for external customer. Ok is a less common for now, but we have to consider it next.

M

Ok, this is a first proposal for the service model. The structure is very simple and followed the one that we used for layer, 2 and layer 2 SM. So there is basically classification between split between access and access parameters and service parameters and please check and if.

F

M

F

M

Can change and we can discuss so finally, in some weird summary, so some operator are involved.

M

G

M

Particular as I mentioned before in the context of a service provider network, could be useful, particular between several department, the one that are more service oriented and the department that are involved in transport layer, and this can be considered Andy scope of. Does he come working group because it's about layer, zero later one technology and another question is maybe we need the layers on to mif for this work and the outer sofa? You can consider this work as a basis for four layer.

G

M

Answer is mode a.

G

Dave sanic Ralph Erickson actually took my question and you trust it there with the map. Should we send the liaison to map I would say absolutely, do you have any authors, co-authors from Nokia, on your document that I did involved with or to map the the I am.

M

G

Driving the work in Memphis from no guy's.

A

G

Is Dave Martin, so you may, if you do have co-authors from Nokia, you may want to get them to reach out to Dave. Okay,.

L

L

Lesson to me, F had been brought a few times this week, not only on there one VPN, but else on lifecycle management, some other work they're doing so from a B perspective. We are looking into it and we are it's going to happen.

N

Severity from Nokia- yes, I- am contributed these drafts, but they were is aware of that. Okay is not appear that, because there is not official documentation from MAF, so I think that a liaison is easily be, is absolutely needed to make aware both of this of the standard body of the activity. Is this absolutely?

N

I

I am this is young and also in order to have a liaison, though this has to be a working group document. Otherwise, there's no point so this.

B

Is this is the first point you're right, and the second thing is that we should wish the ATF. It should be aware of this work going on in em, yes, but only only some companies have access to this document, determined.

G

Let me ask you a question: does it require a working group draft to have the I see camp chairs and the liaison to map asking for their work on layer? One MF should send it here, but given that MEF just had a meeting last week, he's I mean okay, so we can use the informal channels, I'm sure.

G

B

Know what's the problem of the MAF, this document is not available to everyone. This document is not available to everyone, but.

G

My point is this: we can easily use the informal channels to go back and ask MEF to send liaison the C camp since MAF had no idea. C camp was working on this until while they couldn't have because it happened after their meeting right or I, don't think it takes a working group draft document for the two chairs to solicit work from the map. These.

B

Would be great, we rely on your a PR skills.

O

Hey IPF's work is open, so open, right and so I would say if MF participants, you know you know aware of this work here and they think it's interesting to meThe folk. They can just send an email on their lists and inform them or they can access this document. It'll be very it's not for us for us to hear the gossip that map is working on something and then we send them Eleison to ask for that information. You know it's it. You can. Let Neph people know that what it's public right give them.

O

The link say if they're interested.

O

F

O

But that Smurfs yeah exactly.

C

Okay, I have one interesting comment for clarification. So what do you mean they are one here? Do you mean that little wine needs you no audio pipe, or it also includes, for example, of a song.

M

C

A connectivity, audio connectivity or.

C

Because I think people should have their symptom knowledge in mind.

M

C

Because for the young Chui, would you find the optical interface? What is the optical interface? You know actually.

M

We should make some cases about that. This is some examples right.

M

Maybe in the next version we can add an example section where you can make out to use these parameter. Yeah.

C

Right that will be great yeah.

B

I

So next to presentation is a flux grid, optical networks, first topology module.

I

Yeah, this is so basically based on RFC 769 eight and we extend it from generic tea topology model. Next I think this was the previous one, so we can move on to the tree yeah and it is then and then they're compliant yeah. So this is basically a current flex-grip topology model similar to wson model. We cleaned up most of overlapping parts, except you know, flexicord specific.

I

The things are listed in this model.

I

And this is a tunnel model, you know, in other words, to docent a eternal model. We also augmented from teeth on a model and specify only of flexicord, specific terminologies and the parameters here so I think I think it has been a while that is.

I

Draft has been presented in a second working group and and I think the only issue on the flexgrid young I think we study if hold shift, alt p or to be more consistent with the teeth of a model and tunnel model. I think the media channel may not be the best terminology, but the original author wants to use that terminology. So that's one pending issue, but it's not a big issue and another issue is same LTP both model.

I

It doesn't need to use a transponder characteristics, so we're going to look it into to be more consistent with generic model, but.

C

I

I

Yeah, so we think that this is good place to be adopted as a working group. In any comments.

B

So what are the issues that we do? The terminology, because this is the terminology of the framework.

I

Media! No! Yes,.

I

No I, like security, no.

B

Good boy, we have misalignment between the framework and.

J

Please dieter dieter Bela Nokia. First of all, I would like to thank you for incorporating our comments that we provide it at the last meeting. I think this is actually good. I have just one comment regarding the operational the available operational node attribute. This is currently readwrite. I believe this should be read only read-only.

I

You're all right, yeah, you're right. This is read-only. Okay, thanks.

B

Regarding moving forward, as I said before, our priority now is to progress their work on young models for a wfx agreed, OTN microwave. So, okay, it's in it's in the pipe all right.

E

We're going Laura.

B

B

Together, animation.

A

So yeah I think that this third presentation on flexi framework for gmpls control, plane we had this is a port, was version posted and we done a lot to the document and it is basically two things we have been doing. We try to align it with the work going on in itu-t and as that work is also aligned with uif work. We should also be aligned by to the UAF work or the other way around I. Don't know they. They are basically doing the same thing.

A

Then it took away part of the document, especially we took away the use cases. We take cook away use case not because we didn't like them, because they weren't actually focused on the control time. They were focused on the more generic flexi behavior and then we have converged on terminology I think we are pretty well done there. We have done a lot of smaller editorial changes, but you still have a lot of lotta. We have some document clean up to until be to be done.

A

The information we added is mostly responding to to comments we had last time. This is in the comments. Were this is unclear, and this doesn't really capture what you want to do and I think that's now, it's well inside the draft next time. I did show this last time. I all I thought we needed a reference model to actually be able to discuss things and I have asked help from a couple of people were involved in creating the reference model for pw III and MPLS DP factually get all the small knits and bits in place.

A

I, don't have much to say about this next, like this, okay, so EMP RS might be used to set up a number of thing in a flexi context. First of all, we could set up a flexi group.

A

We have one problem where we haven't really converged and that's on the signaling channel.

A

My understanding and I think she lays understanding and in Rada and the thing are not exactly the same, but I think we are slowly come converging and we've actually sorted out. Yeah I, don't see any really problem it. Just like you described will have all how it is. So we can do things through an outer band, signaling channel. If you want, we can always do that for everything.

A

When setting up a flexi clients, we actually have a signaling channel available on the Phi 1 2 on top of channel or one channel available or- and we also have a single channel available on the flexi group we could use either. We need to decide how we actually want how we don't how and in what situation we want to use. This also use those channels, and then we can use flexi to actually announce the PE parameters.

A

We can use the GMP less control want to control plane to announce flexi, TM te parameters in the routing system next and then we can actually also use it to set up and vanilla MPLS LSP and that's actually the example. We have in the in these slides, okay, so flexi configuration an operator could take a number of different approaches. He could either configure the entire infrastructure and leave the Flexi MPLS LSP of FX e to the control plan. Only or you can set up the.

A

Flexy, sexy client with the DMP, less controlled line and only configure the Flex you grow or you can actually do everything on the control plane. The discussion there is: how do we do with a B with the thinking Shannon about it? I think we have it captured now, so it should work.

A

Yeah I think that is nice. So this is we. If you want to establish a flexi group, we have two boxes: the dotted lines. We have a data plane or flexi data plane and ashame on each side, and then we have clicked once we have an NMS that are doing the configuration click one more. You have a routing system that actually is the ultimate recipient of the information and then an MMS yeah. So the NMS tell the two boxes: to establish the the flexi group and then the likes.

A

The note inside send the information to the control time that actually announced it into the routing system. I think this one yeah. So we could also establish the flexi group from from a from the control plane and then the the trick here is that I thought and that's when I wrote did the slides that I actually need the outer band Channel.

A

She lay and Radha has actually taught me that I know I, don't really not. I can use the singly Shannon on the Phi. So as soon as the fire is connected, a signaling channel that I can use so yeah click. One more through the routing system.

A

Is that right, yeah, okay, go on to the content! Allah Allah is centralized controller, for example their own PC box, or whatever tells the control time that you want a you jump too far.

A

So knowing that, when the click when the Flexi group is there, the next step is to set up a flexi client and now it can go ahead. So and then you have a centralized controller, for example pc, but it can be anything and the pc.

A

It tells the control plane that you want the reflected clients on top reflect the control plane in the tuner, communicate with each other and actually are ordering the flexi client to be set over up over the flexi group. So the Purple Line inside the yellow is reflected lion okay.

A

So this is what could be resolved so flexi capable notes, blue squares. You have links between those note, blue links and yellow links. The blue links our flexi capable, and you have all the t parameters and it's possible to find a way for setting up a MPLS LSP. That really requires to be run over flexi links.

A

You can find a way from it and you can actually do the signaling between the notes, based on, for example, next visit er based on any arrow next, so here I again talking about the approaches that could be taking about by an operator, and it depends who you are in the operator shape. So if you own the entire infrastructure and own it all the way up to the the LS piece, then yeah, then you might need the control tend only to be able to set up quick restoration. Otherwise you don't really need it.

A

If you are selling off our selling flexi groups selling flexi clients, then you are in a different situation and you really want to be able to sell flexi clients in a manner where you can actually guarantee that the craxi client you are selling to someone is actually reserved exclusively for that, for that, for our customer customer might set up more than one more than more than one. I will see you that flexi client, but no one else can set it up. So that gives you an unintentional bandwidth.

A

Okay, so that's where we are at the moment, I would like to stress that I think I think we are aligned with other people working on flexi. The draw should be pretty clear at the moment and I think what we would like to do is now at having have a good review. I asked a couple of people that I know dilgar reviews to look at the document to please, as the working group dudes also. We need that, but I don't think we have any any showstopper now to go ahead and accept this working group document.

G

Dave Schoeneck Rob Erickson, um so there was a liaison to see camp I believe it was months ago from pretty.

A

uh Prior to the previous meeting, yeah.

G

From the broadband forum, letting letting C camp know that broadband forum is doing similar work on a reference architecture and nodal requirements, so it would be good for them. There's been a lot of discussion in those groups on what is the applicability of champions to flexi and there's been a lot of confusion.

G

Good to the Asus document. I mean obviously it's public, but it would be good to send a response liaison back to PDF and say: look we think if this gets adopted. Look. We think this is stable. Here's what we think the applicability.

A

Applicability it actually cost protection is one one thing and the other one is actually it you require to run over, but so.

G

Your diagram, where you're, showing the selection at the client layer of flexi capable links, is the first time I've seen that type of architecture. What I've seen in other sessions has been the signaling of flexi, some mysterious flexi information that allows a flexi path to be created through the network at the flexi layer and their question come right. Yes,.

A

G

The question comes in what is the label in that case, and is there something called flexi? Switching and categorically people deny that there's no flexi switching and yet how does one create a flexi path without flexi switching? This is the first time I've seen something that even resembles a path through the network, but it's chain.

A

Of flexi links what we are actually communicating is link capabilities, that the paths require and that's why why we need the routing system.

P

Just one clarification.

P

Here you mentioned I'm, he ariseth. She.

A

Couldn't really.

P

A

To the mic, okay,.

P

Sorry so I just want to ask the flesh you can, inter setup, shaker by request, request a fall, and, yes all right. We should have enough here. You use the MPR as she as you know, that maybe we can use another to carry other traffic. Maybe some I can.

A

Try to answer never see ever gonna study question. So if you are, if you have the Flexi groups established, and you want to set up an MPLS LSP over those nodes, you need to also establish the in between node link on a flexi client lab before you can set up the ambulance. Lsp.

P

You are clear, let me ask a question our way, so does the mg I am just as she here has a stronger. This is safe with fresh, so.

A

That's the I, don't care can.

G

See so I guess the question is: what is the relationship between the LSP that you're talking about setting up here with gmpls and the flexi that set up underneath and how do they relate to one another because they, let me let.

Q

G

Say a common misconception is that you're using flexi you're using gmpls to set up the flexi groups, but that's not what you're doing so.

A

Actually anything that is established in the flexi layer, group or client that is actually announced to the route into the routing system. The routing system knows what the quality all or the capability of the links in network are, and here we actually say that we need te parameters to actually describe the TLX. So when setting up an LSP, you can pick the right type of link.

P

A

Answer to your question: yes,.

P

A

A

So David I think you are required to speak to the mic.

G

Yeah exactly so, instead of say instead of flexi between the two notes, they're using lag right and so at some point, the lag, bundles three links together and then announced so the routing system that this is one aggregated link of some. You know so much bandwidth. It's really no different than what's happening here. The the flexi bundles are being connected together and then are being announced to the routing system as a constraint.

B

Gabrielli dieter and then we have two cattle, okay, Gabriela.

E

Got a meta Francisco: do you think it to a stand, or so the LSP to the case where actually the LSP is a carry door where a DW network, so that your recipe capable interface is become, is connected to an LSP capable transponder? And then you have a uni interface in between is the question so.

A

I'm again, I'm not exactly certain what you are saying.

E

In other words, they Ashima connecting the.

A

E

Capable nodes may be editable the N network. At this point, you need to interface your interface. You know the flexi capable know the interface to a LSP, capable transponders, and then you have a uni signal. You know we're there to negotiate at a bandwidth. I think.

A

The answer is yes, but I'm, not exactly sure the the note we were looking at can take in well an LSP on one side and actually map it down to the to deflect the client. That's I don't see a problem. Okay,.

E

J

Dieter and a Nokia I would like to echo what they've just said. I also get comments from experts from my colleagues who are saying that this draft assumes that there is flex is switching capabilities in the napkin dance.

A

We know it's not room, that is.

J

Factually capable is reaching, and, secondly, I think there is also a locking mechanism defined, which would allow to change the bandwidth of the flexi group, which is also something that's probably not aligned with the flexi implementation agreement.

J

All right, I, don't.

A

Really can I can I can.

B

You repeat: please.

J

There is an administrative locking function defined and you can actually then add or remove files from that group and I think that's not aligned with the implementation agreement of flexi.

J

We can't take that offline yeah.

A

I can take it. Offline I will ever be interested because I, but I thought. You also said that they assume that we do flexi switching. We not definitely not, and it actually were a hint that we could do it in the version 2 of the draft, but it's been out since then, and that's four and a half year.

A

C

Okay, I think more questions can go to the list.

B

Thanks a lot: okay,.

P

Hello, everyone, they say the chilly the present, and this is presentation included to draft one- is the the under 100 Gig framework country, Poland technology and the other one is single, react things.

P

P

During the program meeting first person with them, what the drafter fram-o but afterward discussed- and we also have some questions and some topic of the discussion which what he is using all his intestinal using as cheap Pescara, we have not included a scene in the framework document, so just made an english after I had a new sex into this part of what is this? Is she so?

P

Actually the RSV has a number of end to end the instance founded as a corpus so that his definition so another what I needed to do. I complain. It's asking you to finish the setup with a finish the pounding of this instance as well as indica, some sloths that and may may be unavailable.

P

This is the first string and second wise.

P

The second one is okay. Overall, you saying and after that discussing with things this this can be. This can be solved by reuse, the mechanism defining I, say 71 39, it's because at one configure descending had a most the symphysis I thought UK. So the piano 100k control entirely millage in this traffic to focus on setup of the for her focus on the bonding of the instance.

P

Also, some use case accent I combined the into one category, as they all represent her transporting on out here in Canada, singer, /, auto flex, and then all values in connects the some strikes his job. Okay. Next, okay,.

P

And the following is a singer named after that's what, before this is meeting the other reversal of a single area, tensing all uploaded so next week.

P

First, let's take a look: the extensive, including English justice, our over rail, a new encoding tag, the daughter, seven out, you're, saying it define because in the latest of us amputated of the seminal 9ot you saying actually plays opposite of sex in layer rules instead of a digital passed layer, which is the rule that ot okay, pray also for new pillow area into the existing 2012 was imitated, semi, nine and we are based here.

P

The traffic parameter for audio, saying and also comparing to a FC 71 39. There are also some changes for this. Up to this object. The traffic parameter country will unleash this project. A story is the object that only cow includes a signal type which can be three different single type and the number of instances requested and the Patriot annum I'm. Here they are, they which also try to reuse these two fields with it also in describe the UFC 71, okay, okay.

P

Next one with the final labor, we use a little who indicated instance point in this audiences audio saying with another labor extra included, I'm belogus with his policy in ecology. In his instance RT the pit made plans and Stroud bit a map under the number of available slot and the pretty matalin's and hospital here as a x-ray user that we indicate the prisoner available to us when signaling as user does that have the audio same pass or audio co2 using the same past. Okay, next.

P

Also some other theories that define when they find a new TV available slots class. In theory, which is a new kind of a, must, be attributed it's. This children is expended to collect available throughout the information for and to an audience in ours. It's because one audio CSP miss by multiple what you're saying or what you're saying assembling. Also the interface ID, as we hope object, must be used in reservation message, and it contains several share with three indicator which component in the face you're carrying a specific out. You say how do you see, instance?

P

So, okay, thanks this light, does have some description about how to use g-cloud. She already find in the single knowledge after four actually in the current, addressing of the single image, after which think this is a still mat, Hillier scenario. We can see that it's a case of a setup of a high rack or RSP. It's, because when the example and to end audio serious, becoming Kerala over three different, what you say all the obvious in Tessa and recipe with I can't connect Shakira.

P

So this here does the given some description of how to use a traffic a parameter for requesting the OSP, which have the case of saying same decimal with available slot. You.

B

Have more minute, okay,.

P

Also here with us in some descriptions, a labor how to sell their labor, how to set up a the map, start bitmap. Okay, next.

P

During discussing with also have some open issue, I think we needed to discuss this this week. First, one is audio, say not using it can notice whisper. So why is the so-called and when the RSP, because there is no labor to be switched so I think we need to discuss, is also another issue is how can we define the funding accessing? Can we call this set map of RSP its content now answer when you to discuss this another one?

P

So is there a funding over Oh TOC instance in what you saying is called some houses, saying that the OTO same as he may not need to be explicitly specified that so we we need to solve this problem. I see okay, next next step, asking awaiting needed: okay, because caper discussing the framework and a solution and some alignment more needs to be done between a farm worker. Second, only underwriting draft. Also, we need to discuss where the user case documenting a framework as often so you need to be careful because of them.

P

So many use cases, if not a way. You need to remove that this.

P

Oh, oh also, a terminal moaning. We also the way you meet has at that registration desk at 9:30 a.m. to everyone interested okay, John. Thank you. Next.

B

F

How my name is Troy you found from Maui so as routing extensions, closely related with signal yin and the three more craft. So we just upload this protein stacker merge draft to before just before this hi DF and it's a draft is still in progress and the detail. The technical points are still under discussion. So I will this rap that actually mergers, two drafts about the routing intentions for the AH 100, 100, Giga OTN and all the the extensions actually mainly focus on the LCD part?

F

So for the first drafts and it's this draft has been presented in previous ITF meeting so in this drop, the type 3 bandwidth sub T are, we is defined for the audio see in container and at the TSG? Is that you send it for 5g granularity? The unreserved bandwidth is not included as audio Shion is not switchable, and this is the detailed of the label introduced in the first craft, and they is also present. It has have been, has been presented in previous IDE idea and for the second draft.

F

A second draft stands the type 1 bandwidth sub T. Are we for ot? You see for audio C in container and as audio signal type is, has only only has one granularity which is 5g, so the PSG is not extended and the number of T is used to indicate the number of available available tributaries rot, which means the tributary slots can be occupied. It's different from the can be allocated how to say there are two okay. This is ambiguous.

F

Actually, this term res, they are, as there are two pieces in the boat for the voyage showed you same time. Velocities can be occupied or it can be marked as available. So this is so. This is used for the availability and so in the in the merge draft.

F

Actually, her draft merge the two, the two drafts, and so the technical point for the leap, the ICD part is actually combined and the two two drafts and defines a type series up and we sub T RV and extend the 5g for granularity, as there may be more granularity, is for the audio CSO and a bitmap is used to indicate the exact locations of the unavailable tributaries throughout the information. Although.

C

B

L

F

A Chile has introduced for in the signaling.

G

F

Although it's the you know, unavailable TS information is described in the signaling part, but actually I think this is. This. Information is actually is up about the node capability so which is which should be carried in the routing advertisement, so it is added in the merge, straps, and so here is the difference compared comparation of the difference for three drops.

F

We can see that the merge, actually the merge raft, can combine the two two drafts: first, first of the RSP encoding time, audio CN and bandwidth, sub, T, AVI type and you standed tributaries South, Korean variety and the fire, our reserve bandwidth for the audio Seon connection, setup and the information about available tributaries loss, and so next I thought we. We are open to discuss and revise the technical detail for the merged routing drafts and any comments welcome. Thank you.

B

Comments questions I have a quick one, which is, since you are defining a new switching capability. Please consider using the generalized SCSI general access and okay.

F

Okay, we will try to Al Kut, consider this okay. Thank you. Next.

H

Hello, it's a humming from highway again I'm going to present this work talking about the interworking of the GMP, RS control and the centralized controller system. So we are coming back to GM PRS again and firstly, I would like to introduce the motivation of this work because we have been working on the GMP RS as a control plane technique for a couple of years and it before the the controller emerged. The data plan is consistent. It's working consistently with with the control plane, but nowadays with the involvement of the controller system.

H

We have also developed a kind of a pressure of solution that how the controller system interacted with the corresponding device. So it would be interesting if we can further understand okay, how the kind of distributing the flavor of solutions like GMP RS and the centralized the flavor of a system like controller system, they can coexist and they'll work together. So the example protocols in the GM, Paris control pain may include the rsvp-te protocol for signal and the OSPF t for routing and correspondingly there has been a few newly emerged.

H

The technique like rest comp protocol and the young modo defined for a kind of centralized device configuration, and we also have a kind of a PCE protocol to use as kind of centralized routing or pass computation, and the objective objectives of this draft can be divided into three different issues: the.

H

Firstly, we would like to check the current architecture from different the functionality that we make sure that the current architecture satisfy all of the requirement, and we would like to understand how the existing solutions can be applicable to these kind of architectures and during this procedure, would believe that there can be a kind of potential gap analysis and, for example, we show an example in the GMP RS, together with a city in the architecture and, as you can show, as you can see in this figure, this is a kind of multi when there multi-domain stuff and we we have GMP RS as distributed control, plane technique.

H

Well, we also have the rest compound and the young model as MPI solution, so there we need a kind of consistency and the neat approach to make this two system a inter work together, and in this page we try to list a few different.

H

Consideration on how this in her work with each other, for example, on the southbound interface between the PNC and the interface, how the GMP RS can help report the topology. We may have different alternatives like PCRs from the centralized style or we can have a kind of OSP FTE in the distributed function, and the second issue would be like considering the inter domain link and the SE need to gather information.

H

But usually it is a kind of stitching from the information reported from PNC and how the JMP RS can represent this kind of information to PNC. It would be a key approach to this problem and correspondingly in the Nets, the issue would be interaction, request for the RSVP because early giving a kind of PC protocol. If we initiate the the past supplement from the from the hand had a node, it would run the rsvp-te signaling accordingly, but actually given multi-domain case. There would not be explicit.

H

We are all function to describe each node in the other domain, so that would be taking care from the other kind of controllers. And finally, the monster advanced should be the kind of protection and recovery. This would also request a very accurate interworking mackensen between the device level and the controller level. Some kind of the intro domain protection and recovery is very different from the inter domain operations.

H

So here we are trying to introduce some issue and examples for our interworking consideration and in the next step we are going to bring more details and describe some typical interworking scenarios and during the procedure with, if we detect any kind of gap, we would like to continue our working on extending those kind of protocols. And but before doing that, we would like to confirm with the working group that we are doing the right job to say that this work is useful.

H

J

C

J

R

J

Actually, I I think this is a good piece of work because it shows actually how gmpls networks could actually interact with the actn components. Can you go back to the previous slide? I think I can answer easily a question number one.

J

The PNC, of course of each domain could actually listen to OSPF te and then it gets the topology yeah.

H

I think it is not difficult to have one answer, but it is really hard for different people to have the same understanding on how to solve it right our kind of problem. So it may be clear to give better description of how this work is done.

B

I agree with Peter on the usefulness of of this work. I am again wondering what is the relationship with the transport and bi design team? It's probably something that fits in there. I.

H

Think it to to go a little bit beyond the MBI or the MPI in a city, because it is more device level, interaction.

G

H

Some device attributes may be taking into account.

K

In the design team, we are looking more about what empower MDS EMPAC interact each other. We do not care because we are looking at the MPI and and it interoperability space how the PNC translates the information that the mdac provides into action in the network, also because in the network you can have GPS or whatever you like, so I think this second also. The point is useful because it complements is say when you get this information and you are running jumping us what the PNC can do. That's useful information. This is GM.

K

Place is one of the major SPIE protocol, especially defined here. Thank you.

I

Only way oh yeah I think this is a very useful work. My question, the question is: are you also scoping piece up within this picture? When you say gmpls, do you also include he said or not? Well,.

H

Currently, currently I categorize the the piece app as the part of centralized the past computation mechanism, but GMP ours is a kind of distribute it was but you're right. They are highly associated with each other on the spi right, yeah.

I

So my question is: maybe working with GC working group together with C camp, to figure out little bit bigger scope of alternatives because he said you know, centralized method is now if I understand based on offline discussion is going to take off. So this is very useful work, but in terms of scoping right. So if you consider.

H

That that would be single, that's dependent on the result of the gap analysis. We.

B

We have a comment from from giuliana. He says it seems to me that there is much of a lapper with existing PC documents like for the 6:55 thanks young agree with you.

N

By just a short comment, I said that I supported is that his work and it Isis, is very interesting because for the first time in stride to the findings are working between gmpls and that's the end, control and also open the door on they need that. What mentioned by eternal now, yes, then transfer design team. We just analyze, basically MPI interface, but these thereafter make care that I also work on the SBI I. Think ITF as to do it is important to do.

Q

Yeah, who do you consider Telecom same direction as Julian just mentioned, so what kind of problem do we want to solve here with this solution compared to the existing solutions that we have with PCE? So if it's going more to Sdn, so then PCE and napkins are almost not the right mechanisms to address this here. So then I 2 RS, maybe is the best, is more or less a way to move forward.

Q

B

Said that Annette convent PC are not as DM protocols.

Q

Depends on the definition of s the end.

Q

Q

Just to answer your point again so for me, Sdn is programming the forwarding plane of a network, and you.

B

Can do it with net cough and PC and bicep.

E

Okay, this gabriella got in Bertie from Cisco and I would like to talk a little bit about the possibility to extend the current signaling on the SS wanna, also to cover the carrier part and the interface, the client and network interface. Next, please.

E

So, even if we have let's say pretty excellent 76 99 98, sorry that is defining the frame of 40 sson and also 6699 and 77 92, to define that the signal or a media channel, we don't have actually the possibility in case we want, to, let's say, define and signal what is inside the video channel between the two clients. Next next again again, please. Okay, here is a picture that described a little bit the issue.

E

So, supposing that we have image node that has multiple DWDM interface or SS, one interface connected to the client to the core node and then suppose that the edge node wants to ask for a media channel to connect the tail end. Node. So, basically, okay, gmpls can calculated the media channel can define the path, can define the N and M value, but cannot or if it is able to do to define.

E

The frequency of the different sub carrier then has to signal to the edge node this frequency next and on the other side that the edge node. How can ask to the corn order? What kind of circuit wants, for example, can define number of that can be fixed or maximum and I can define the total bandwidth? For example, over ten for carriers, the H node can ask for a foreign Turkic Channel, or maybe an 800 DP in a tendril gig channels, and then let the gmpls calculate how to implement that I mean.

E

For example, foreigner can be implemented with the four channels for carriers planning at 100 Giga and the specific modulation format, while can be as well calculated and implemented, using just two subcarrier running 200 geek and also again, a different modulation format.

E

Great on the other side, when the gmpls calculates everything as to tell back the edge note about the modulation format at the FEC, the baud rate and also the position of the subcarrier is added with a channel, so J and K are reported to next slide, a reported n related to the media, channel, Center frequency and, of course, the media channel. With on top of that extra this, we have defined also the relationship between the carrier and the port, and also the carrier power very quickly.

E

I think we address a real issue here, because we realize right away during the implementation, this sson, that something was missing and I would say. We are available to explain in detail why something is missing and how the real issue is addressed here.

E

So I would really encourage to either meet in person or run on the mailing list, any comments, and they will ask thanks a lot.

J

Nokia I think we don't have to just reduce custom. What we have discussed in the past I do have reservations. I know this is an experimental draft. On the other hand, this is specifying a proprietary solution. So that's it's just for the records. Thank.

E

You thank City that now really we can I encourage you to to discuss what are we we are dressing here, because, even if you don't have yet this kind of problem, you will have shown don't do it to solve thanks a lot.

Q

So this presentation addresses.

Q

Two drafts the first one I just want to explicit I mentioned that Nokia and dieter is, is only supporting the first document here. Yeah, okay,.

Q

So, based on the framework document dealing with the optical interfaces that are not part of the managed domain of the optical network, so managed by a different network management. For example, we are using LMP and the extensions of Olympian this both documents to exchange the properties of one side, the optical line system and, on the other side, the interface itself. We defined certain or different sub objects based on RFC 4209 here and these documents, so, first of all to negotiate the capabilities of the line system and of the transponders.

Q

So, therefore, we added the application codes and the application identifier x'. We are supporting as well a super set of parameters that are defined. The current specifications of itu-t I have to mention the first document that we discussing here for the single channel optical interfaces is on standard stroke. The second one is an experimental draft for the second one. We addressing the multi carrier. Optical interfaces for spectrum switched optical networks, so some things we have to exchange or DMT.

Q

The parameters again not for configuration, that's very important, just to figure out which capabilities we have on both sides. Next page yeah, I think for the current version. So we updated both documents for this ITF meeting. We did a little bit of bug, fixing not so much things reshape the documents a little bit and added some parameters on the on the draft for the Flex grid. Next page, you can jump over this.

A

Q

Additionally, to the the application identifiers, our application codes, we added the the output power and the current input power for the single channel interface document. So that's one thing we have to mention here and next page. On the other side, a little bit more parameters are needed to exchange. So all the details are in the in the draft. You can read it and give feedback to that next, page yeah and that's the last one.

Q

So we want to request feedback from the working group and we think at least the single channel interface draft is ready to for working group adoption. Okay,.

J

Dieter Nokia I have a question regarding the new parameters that you introduced: the max Paul power difference and the max pulse q. Difference I didn't find definitions of those our references to definitions of those parameters.

E

You will probably see in the next I to t69 8.2, that is coming.

B

Any more questions: okay, there is another comment from Juliana, but I think it's. It refers to the previous to the previous presentation. The comment is experimental, but the target will see comfort, not the individuals, treatment, I. Think it's a counter commenter to yours, Julian.

B

Okay, we are ready for the last presentation.

S

Yes, hello, everybody. Do you hear me? Okay, yes, okay, great! Thank you yeah. So this is a little bit another kind of module. This is a generic alarm module that lets devices. Press enter alarm state. A first question, it's hard for me to see which of you will raise your hand, but how many of you have read the module at all a.

J

Couple of words, some.

S

Yet yet I will keep this fairly high level. Don't expect please.

S

S

Okay, take the next one yeah. There.

B

Is a little bit of delay if you.

S

Tell that they.

B

S

This one, this one is okay, so it's been more American that is working on this. One I would like also to thank there's been relevant input already. I had one this morning from Balor's from Eric song, I've had good input from Joe and Nick from atra and also low barrier for lab, and so you'd iterations on email list. Already, the main features of the module is a way to represent alarm list, which means what are the alarms on the system.

S

A couple of things here is that you, if there is a clear, the lower meaning from a resource point of view, it's cleared, it still stays in alarm list. There are other definitions of alarm lists out there, which behave differently.

S

There is also an yang feature to support operator actions on alarms. The classical ones looks like acknowledging.

S

There is what I think a very important thing, which is sometimes lacking, which I call the alarm inventory a way for the system to publish the possible alarms today. It might be the case that you get a document from the vendor in the free format and then some kind of protocol, but this module lets.

S

You ask the system, which device, which kind of alarms will you actually publish, and there is also a way to block and filter, alarms, shelving you reusing a term from the process industry, which some degree is a little bit more amateur in when it comes to arms that then, then we are in IP and tell Co sometimes, and you can also do some administrative actions on alarms like deleting, alarms and then purging the history.

S

In some cases you might have a more tell Korean debt management system and there's also a separate module here which lets you map the alarms into the classical telco standard, XML III, but note well. This is a separate module. It's not part of the core module at all next slide. Please.

S

Kit so when we say alarm in this module, we are referring to an alarm as a state on the resource, not a notification. So conceptually an alarm is some undesirable state in a resource that requires corrective action, meaning this is not for events in general. This is not for logging in general. Every alarm is important and it needs someone to take an action. A human or a software, but the meaning is don't use this to throw just events into it's serious things that someone needs to take action on.

S

If, if that's not true, it shouldn't go into this alarm module, it should go into loading somewhere. That's the first thing, and the second thing to note is that we take a stateful approach. When we talked about alarms the alarm is built on, you have Minyard.

S

So what's our question.

S

Okay yeah. So if you look at the list to the left here when we say alarm here we mean a resource like a specific interface. Oh sorry, go back was more again.

S

Go back to the previous slide. That's perfect! Thank you. What is an alarm so when we say an alarm, it's a specific resource could be a specific interface, a specific disk or a specific detector or something and then a specific type of alarm. So those two together are the key in the alarm list. It has a severity and a separate clearance flag. So if there are some alarm geeks in the room here, some alarms stand and sort of confused severity levels with clearance, East alarm cleared or not. That's one thing.

S

The severity of it is another thing, and so we have a list of the resources. There learn types their current severity and if you're cleared or not that's the focus of the alarm module and a society effect, they will emit net cough notifications, where some modules more stays on the right. Focusing on the notifications themselves. We focus on the states themselves next slide. Please.

S

Good, so here we see the alarm list itself. We will focus on a couple of things again repeating what I just said. You can view the alarm list as a function so from a specific resource, a specific alarm type. What's the current alarm state. So that's the purpose of the alarm is it's a function of that kind? It's not a notification. Log and again clay is separate from severity. So what you see here to the right? You have a read-only alarm list. The keys are resource.

S

The interface, for example, the alarm type, the specific kind of alarm I will return to that and those are the keys, that's important. So the time is not the key. The resource in the Laurentide are the keys, and you can see further down that there is a separate field, saying is cleared. Is it cleared or not, which is separate from the current severity level, which is perceived severity then for each arm? You can have a history of State. This changes from the underlying resources.

S

So if an alarm might go from warning to minor to cleared to major again so this date rei.

C

7 move forward quicker, yeah.

S

I'm coming today- and this is the important thing so this will actually be my last slide. The other slides are for reference, so this is the core thing. Nope, please go up again, previous light that it will be my last slide, because that's the important thing now up up again stay on the list.

S

So there yes stay on this one. This is my last slide, so that's the state changes you can see the the history. What has happened with alarm from resource point of view, then separate that further down this list called operate to stay in change, meaning an operator might have acknowledged alarm. I. Tell you see. Little arm is being closed, that's separate list, so there is no thing like an operator clearing alarm, but that's two different perspectives.

S

So this is the alarm list and that's the heart of it and notifications will be emitted from this one and the other mechanisms you can read it RFC you can block and filter alarms. You can query the system. What are the possible alarms that can be emitted so that that's the main thing around this module? So any questions.

B

Okay, there is no question from the floor, not just a consideration from my side. Well, actually, I'm happy today that the draft generated a lot of discussion on the on the mini list. This means that there is there is interest. There is also rather support. I would say, since most of the comments were what positive I didn't see any next step in your in your presentation, but I, guess you you want the draft to become a working group document sooner.

S

B

Later is correct, that's correct.

S

And I have good input, requirements and comments on the email list will be, which will be worked in an updated version, mainly problem from the a drink eyes and from there excellent yeah.

B

Okay, I think this is it. Thank you very much stephannie for for the presentation, thanks everyone for being so fast and helping us the closing the the meeting almost in time and see you in London. Thank you.