Internet Engineering Task Force 95, 6 Apr 2016

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: IETF95-TSVWG-20160406-1620

Description

TSVWG meeting session at IETF95
2016/04/06 1620

A

A

A

B

Would suck some doesn't touch is the point. This is TS VW g, if you're hoping for something else, then stay anyway.

B

It's going to be chaired by me and Martin David isn't able to make. The meeting will hopefully have David back for the next meeting is the audio working.

C

B

Go and get some more people.

B

This is the ITF not well, and if you haven't read this, please read it all. Contributions to the working group today will be covered by the IETF, not well. If you don't intend to do that, just leave the room join those people outside, but please read it agenda for today, and the agenda has been bashed slightly and so we'll just go through the agenda. We've got the ecn. Encapsulation guidelines by Bob is Bob.

B

You well, ok, diffserv and I trip, really 802 11 and another presentation requested by Bob on l4s and its use of cold points, we're not having a presentation on tunnel ecn. The authors aren't present at this meeting and they didn't think it was worth doing a remote presentation, so we'll have not yet from that later and then. Finally, we're going to look at TTP, alternative back off with ecn is this draft was previously discussed in TC p.m. and IC c RG, and it's here because this might be the correct home for actually progressing the draft.

B

So this is the current agenda. Does anybody have comments on the agenda.

B

Whoo Tim's doing the noise reduction that should help.

B

We don't have Bob.

B

Fred, do you like to do the first top.

B

D

B

D

B

D

This slide admit.

B

I was different: ism, yeah, okay,.

D

Next slide, please.

D

Okay, so at the last IETF meeting um we went through the services that are discussed in RFC, 45 94 and in this draft, which basically duplicates those services and had a fair discussion and walked away, basically, understanding that the working group was saying that they were fine with the services that we offered and then the chair of rich working group. Does she chair it? The chair of the right working group, whatever was came up and said?

D

Oh by the way you didn't reference, my RFC or our RSC, and by the way our deserve good points and services were all different. So could you please do something about that? So Tim and I went in and did a fair bit of analysis looked at looked at the bad RFC pretty hard and she's right. These services are all different. It's an overlapping set of good points. Normalizing. The thing in this specification is just plain hard.

D

So what we wound up doing was we inserted a paragraph, but you can read here if you have microscope that says in a nutshell: yes, there is this other RFC. We don't know how to know moyes, and so we think that a user of this RFC and of RFC 45 94 and the entire diffserv set of services that it's based on ah that needs to talk to a GSM. A system is going to have to do dscp translation at that boundary, which is described in the diff, serve architecture.

D

It's how you handle different disks or domains and sorry they're different deserve domains, and you translate at that point and did I say anything more there. um No, that's literally what I said and that's the end of my presentation so.

D

So my question for I have two questions for the working group. One of them is: is that okay? Are you guys, okay with that outcome, with respect to RFC 7561?

D

Yes, no indifferent. We.

B

Can take comments in anybody's got comments? We can also, of course- and we.

D

Can probably hear.

B

For the people that are here, we could probably hum um what are my.

E

Questions on what are.

B

You, okay with that or on this well yeah! Okay! If we need people to comment, if they've looked at this yeah has anybody looked.

A

B

This and have opinions on this particular paragraph. The drafts been around for a while and we just added an extra paragraph, so I guess we need eyeballs on it and yeah. So obviously, a welcoming comment will take comments on the on the list of people. Don't okay,.

D

It's not fine that works for me now. Do you want to go through the working group draft question or wait for the comments.

B

No I want to go whether we wish to adopt this okay.

D

So second question is: we want to adopt.

B

This does anybody want to talk about the adoption question. Anyone comes to the mic, they can say things if not we're going to take a hug. You're allowed bring eggs to my throat.

D

B

Okay right and we're going to take a hung, em, please and we're going to ask for two hums and the first home will be pleased hum when I say, if you think we should adopt this as a tsv WG working item and the draft is not on the list.

B

So if you want the 802 11 nothing's for diffserv adopters are working on your part and please some now.

B

Okay, I hear hums, please hum when I asked, if you think this should not be adopted the working I by this working group at this time. For any reason, if you have any reason not to adopt it, please hum now I hear no hums against adopting this. Therefore, I shall take this to the working group list with the recommendation that we consider adoption, the document will yeah and the document will then require reviewers so because the this documents being a simple mapping which we've talked about before so well we're going to call for adoption.

B

We're also going to ask for people who might be willing to review this.

B

B

Did anyone see Baba in the Carter okay, so Bob mrs. is to meeting slots at the moment and.

B

Next up will be Abe.

B

This draft has been presented in before, and it's brought here because it was discussion with our IDs, about where the best place might be to form a discussion on this draft and since it's an update to RSC 3168, which is a RFC maintained by this working group, they invited the authors to come here, I'm an author but I'm just sharing at the moment my Michael's going to the talk over to you Michael. Thank you. Okay,.

F

Yeah, so the context of that is the following sentence from 3168: it gives context. It says the dance system should react to congestion at most once per window of data, most most once per round trip time. So it's not something we're criticizing or changing or anything. The point is that this is a specific contents that context that rrc 3168 sets. There are other proposals in the space of using easy n, and this is not about them it's about this context.

F

Next, there is the problematic text. There is a well the two separate paragraphs in two different places. Second, one is easy to miss. We missed it in the previous version of the draft, also in our conversation so but I've seen it before and the text. The first text says that the indication of congestion should be treated just as a congestion loss in a non easy, incapable tcp, and this means having the congestion window.

F

That's using a lowercase should, but it's it's seems to be a clear recommendation, but there is also another text that says upon the receipt of an easy, incapable transport of a single seed packet, the buyer, such a transport. The concessions were algorithm, follow that the ancestors must be essentially the same as the congestion which corresponds to a single drop packets. Essentially, the same I think from the other context, is clear that this means having, for example, for easy and capable tcp the sources source TTP's required to have its congestion windows.

F

Those explained here for any window of data containing either packet drop or an easy on indication. Next,.

F

So the issue here is that this rule creates poor link utilization whenever we use smoke use when you ecn mark at a point in the queue or you know, even if it's the same point as loss, when you do that at the point that is way below the pencil a product, you could use small utilization with this, in particular with modern hmm algorithms.

F

I think this is not what we want, and this rule prevents us from utilizing the knowledge that ecn gives us, which is that we actually have an indicator here, that the queue is not necessarily a panel today product. So therefore, we think this rule is harmful next, so there was a Dadaist. Is this draft that Corey has already briefly introduced? It was the reason most recent version is called graph. Cadamia alternate back off easy on 03. This draft was presented in TC p.m. in ITF program Yokohama. It was done later discussed in icc edgy.

F

It changes to send a reaction to easy n and it proposes new language for this RFC 3168. This combination turned out not to be good, so we plan as a next step, step to split the draft into a PS update to our C 3168 that just fixes these statements that I showed you and the short draft to recommend an appropriate congestion response. Now the first thing is underlined. This means this come with this talk here is just about this first draft next.

F

So the question I'm asking is whether this working group will be in will whether there is energy and interest in this working group. In.

A

F

In adopting an up kind up, coming post standard draft that will just update the text of RFC 3168 and to no longer recommend that you would have liked, you would react just like you would recollect to loss when you get an easy n mark questions commands.

B

G

Oprah's cohen, apologies for being late cuz. I think I've missed the slot. I was going to be talking in, but answering your question.

G

Would you be opposed to adding a bit more stuff in here to update our esli 3168 as well? I know I've been talking with praveen about some of the control packets that have to not be easy and capable, and things like that and and now there's some work needs to be done before that they were allowed to be easy and capable, but it may be better to pile more things into one update and have lots of little updates. I don't know, but then it would hold hand back your one I.

F

Guess my answer depends on how much it is and what it is, and we were kids per case not not in principle of all.

H

B

Everything yeah and I'll put my chair hats on rather than my author Hut. For that one and just say that if the working group or adopts this, then of course the Wiccan group will have consensus about what goes in. But we don't really want to add feature after feature to this. Eventually that documents going to have to be updated and probably abyss issued, and so we want to try and keep it well focused. But if there's something pressing that the wicket really feel should be changed because it's wrong, then we should do all the corrections.

I

Yeah 3168, if you went 61 9847, um don't want hi, uh Brian, Trammell um yeah, um just a clarifying question. I want to make sure I understand exactly what you're trying to do with that. Ps update you're, basically just removing those restrictions, so is it so that other things can be done in the envy, informational or experimental the other document? Yes, what's your your plan for the other document is informational at this point or experimental, not.

F

Informational experimental probably go.

I

um Yes, then, do it I think it's pretty easy to get done. I would be um skeptical of trying to say, oh well, we only get to update 3168 every and moons and we have to throw all of the stuff in there. This is not an ietf process. Document I think it's a pretty easy slam dunk to just get the thing done um so I'm, definitely in support of okay thanks. Alright, thank you.

J

Eric not mark this sounds useful, but I didn't quite understand. Why need to split them in two separate documents, as opposed to having the PS update update the PS, as well as having non-normative test text saying? Well, here's a possible algorithm that satisfies to both constraints.

F

We will aspire advice from the 80s arca.

J

F

There's one standing behind you, he connects ok, you can probably say whatever you think. Is it good or bad idea? Why? Because.

J

There's an implementer it's easier. If it's one document that says, oh yeah, you know go find it one things you won't blow right and.

B

I wrecked you just to be just to be clear on on the document contents before our Allah yo decomposes, what he's going to say and the the practice a parameter, not an algorithm, so the update it proposes to allow a range of parameters and then the specific one is to choose a parameter which they think is safe. A number a single.

K

So managed managed emitting you're ready for the next two hours and then I'm going anywhere now what the process wise is better. If you have had two documents, because one is proposing to change too easy and the other one is meant to be experimentally. The first place to figure out if it really works on the white squares, attended and now I'm Dave Ted as a designer proxy for him because he's in Jabbar- and he was writing in Java- that he would like lost class ii CN markets in one RCT to be discussed in this document.

F

Morrison marks in one or two teas. That was sorry. What was it more? I c'n maximum velocity, okay, okay, but that's not the just above, is suggesting to change. That's different yeah.

C

Mario, I wonder if there is you know, experimentation to make it proposed standard immediately. I, don't.

F

Understand that the store, could you speak up I, wonder.

C

If there is enough experimentation to make it proposed standard immediately, if.

F

There is enough experimentation to make a pee hole with other values than 0.5. Is that where you are I mean.

C

There are several things that may go wrong. First of all, the when you are now using fly size as we are you asleep or lost base.

C

It works correctly when you an inconsistent avoidance, but when you are in fast recovery there after the first round review suit, still reduce the consistent window. If you get the easy and a call right, yeah.

F

C

During the first recorded flight, size is a random number. It doesn't reflect correct idea, recurrent bite size. That's one thing: oh, the 30 3168 was very careful making this correct, because when you are using consciousness, wind enduring the perfect cover everything works: fine, okay and second, the second in east europe, where you have small window to so sub 6 MSS window. You are practically not necessarily reducing every round window. When you get a sec, an echo because minus fifty fifty percent doesn't mean necessarily full MSS.

C

You practice it won't be produced in your window. Each round trip, as you should can.

B

I get clarification. What you d are you saying there should be a reduction of at least one MSS? uh Is that lot of something like that? I believe.

C

So there are things like that which deserve some experimentation before going TS in my own and.

B

You think you can get experimental evidence for the, so you think we have to use our engineering judgment, but is that is the reduction by one ms s, something that we can do experimentally or something for which really need engineering judgment. So I can see the logic of discussing this. I believe.

C

You can do it. Okay,.

B

So so there are two ways: we pick a algorithm one because we judge it to be safe and the other, because we can measure it and we can know that it is safe.

B

It seems that there might be at least a case for discussing a one, MSS reduction and as an engineering this, because we consider as a working group that that's a safe thing to do without necessarily doing an experimentation to confirm it, because we understand how the congestion behavior works. So I'm asking. Did you feel that to verify that choice, you would actually need experimental results or whether we perhaps could make engineering decisions? What was your feeling more.

C

I'm, not quite sure but yeah, possibly engineering is your son is enough hmm because at least that I believe I I can figure out what what happens, if you do it in a correct way, but there are certain cases that that should be very carefully design. So the context is not enough. That's the point. Yeah.

L

Okay, you can even I have a question. Is this? What your proposed is a specific to the reduction to enter the reaction too easy n, or what this also apply to a lost signal.

C

Partly yes, so so this is what we reaction with small windows. It's the same thing with the loss paste, but you're the first recover. It's different, of course, based recovery handles this correctly. There is no problem.

B

So, okay, this is one of the good things about bringing things to the ITF. You can get discussion on the topics and it's good to receive comments, and this is a chair speaking still.

B

What we need to do is decide whether this is a correct hope for doing this activity and whether whether we should do it so is there anyone who wants to speak on again on whether we should do this as an activity or whether we should avoid doing this.

G

But busca, just just going back to the history of when I, actually 3168 was written and why it was written like it was at the time it was.

G

There was great concern that if different TCP responded to eating differently from the loss, but it it would create a safety problem, if you like, you know that there would be people stretching the numbers further and further if you like, and so I mean I'm fairly well known, for not worrying too much about rape, fairness, but the difficulty of getting together some text that allows us to allow a bit of leeway, but not too much he's going to be difficult, was all I'm saying you know if it's it and the idea of just having one number I'm, not sure it's the right idea, cuz that that's a bit too specific!

G

You know he's trying to summarize that. That's what the bottom.

B

Line this slide says then, which is find some text to set some bones: the appropriate relaxation, but don't set the number in this PS all right yeah- is that what you're saying is: okay,.

G

um As a process, I'm not saying anything's, okay or not, okay, I'm just saying in the thought how you would write this text in such a way that it would be clear enough and you can find enough, but also flexible enough. You know you.

B

Know that's why we have working groups to do this work. It would take more comments. While we're taking comment. I'll make a procedural note. We can't adopt a draft that isn't actually on the table, so we want you to be doing an adoption call today, so we're just doing a discussion prior to in the future and you draft appearing and then adoption column, but we have a working group meeting here where a massive opinion. So please come to the back. You have opinions, so we get to be picky in the worker meeting. Okay,.

D

So I do have one question and that is that I believe that IETF 94 you had what amounted to a barb off and we're discussing an alternative back off proposal and um I walked away from that I I wasn't actually able to attend. I had a conflicting thing, but I walked away from it with the idea that you had some intellectual property.

D

Do you well, somebody has some intellectual property. Somebody has that is there and what? What is the status of that.

D

B

Spin tonight yep I'm.

G

Not sure why it's relevant to this, but the status is it's not being declared yet.

D

It's not undeclared yet, and that's actually, okay, I assume a declaration is coming at some point, but is that going to be an issue? Where are we going to be tripping over ourselves for that? No.

G

Not an issue for what Michaels trying to do here at all.

I

Actually Michael, can you go to the next slide? Next cuz we have the actual practice.

B

Out the previous slide, Oh.

I

Is there, then we have.

B

I

Yeah they dropped on top is an extra slide. Yeah.

B

I

Now, just just showed up real quick, so there's ooh text and make it go away because I actually yeah cuz, so I've actually had a look at it and um then what but yeah I don't know. No I, really really really don't. If he wants it, I really, don't I really don't want to have the debate about this text yet I want to say that we're gonna adopt it first. I want to say now that I've actually seen the text. This is a good start. I think we can quibble about the numbers a little bit.

I

um I'd want to talk to you offline about, like where you know where the numbers came from I. Think there's number equivalent to be done here, but that should happen within the working group and unless the new area directors are going to make a reorganizational change to where the ecn stuff lives, then this would be the working report right. So this is the ecn, maintenance and minor extent. Working group is traditionally the ecn. We can Grissom ok trying to answer both questions so.

B

No, that's fine, Brian and I'm keen on people quibbling with stuff and making comments here, because it indicates some energy to actually do the work that needs to be done to get that text just right. So that's good you're me to start.

I

B

B

L

Make you live in turn also, I'm I'm not super sure how the text should look like this. This is not cool. I.

H

L

Want it go into details, but MC I'm, just saying it's not not that easy to get it right, not sure. Alright, we can get it yes, I agree to jump up here and the other question is: do we need to update the draft and I agree that what's written in the draft is extremely confusing and should not have been written that way, but I'm still don't know if you need to update the draft, because it's still written in a way where it actually kind of could be okay to do what.

B

We wanna do it update which dropped the.

L

Sorry not drive the RC, ah ok.

B

L

E

L

13 yo's now.

E

You have to be difficult to get around the wrong side. Take the other one, oh yeah, and which.

B

One, do you want this yeah.

E

L

Nick sonyae, so I didn't want to be picky, but this is essentially right and just you know changing your number from 0 point 5 to point season. 7 might still essentially be the same reaction or whatever I don't know. It's like yeah.

B

Well, okay and I.

L

E

L

This as Nikki I'm, just saying I'm just saying I'm I understand that it would be nice to clean it up, but I'm not sure how correct we can get it. And so I'm not sure if the update is needed, but I'm just I'm, just I'm sure.

B

That's mirror I'm sure, ok, so right, ok, so now I have to think as a working group sure what I'm recommending what I'm recommending is. At least people have opinions on this, which is good news and I'm I.

B

Would personally encourage that we try and find a a document that clarifies this point because we're going to be working on ecn in various forms as things go forward, we keep advocating the use of ecn. I've always had a lurking concern that if we do this, then we have to be clear about what 3168 says so I would encourage you to resubmit the draft and then go to the working group and ask if they can adopt it here as a work item to work upon and if you're doing that.

B

Please do split at the second document, because we should be discussing the number while we're trying to get that. We could make two documents bring it to the working group and then let's talk about it.

L

So that's actually a good thing. I think if we can come up with the text that actually everybody can agree to, and that totally makes sense at the end, then why not updating a right but like even even and I I think we should work on this. So that's something I'm sure and the point is even if we adopt the working the document and then we cannot come to a text that actually makes sense, we don't have to update it. It's like only if we actually can find something that makes sense. We have.

B

No because an ad problem, then at some point we have to decide what do with the work item. But for the moment that's the planet like that. Let this place make two drafts and bring it to the working group receives as energy yeah.

L

In this case, I would like to have more discussion before we add a milestone. You.

B

Would like one more.

L

Discussion on the extra text before we atomized stone.

B

Is that mirror as mirror or.

L

Is that Marisa support your daughter? It's my personal opinion here.

B

I'm not sure what that means. It was totally confused.

K

So I would say just just read the draft submitted for the group and figure out what's going, and then we can see how it ends up. It.

L

Was just because of Corey was saying the word and wise don't it was welcome, what's kind of like just just.

K

Do it all right.

B

Okay, so you have an action when King do it for Michael? When can you do it for when you're going to revise the draft I always ask this ah I mean, as is it in the next.

F

H

Something like that is that good enough, this is.

F

H

Give me another one too cool once Jeff time. Thank you. 1.5.

B

Right, fine, okay and so we've reached the end of our agenda, but we didn't. We didn't get Bob schedule, so we need to go back and showed your Bob Bob. Would you like to talk about the EC entendre, no easy and over there too.

B

Which is this? That's fine. We.

B

All right, but.

G

This is a sub turnin around.

G

G

Hello, oh, is it ok Gary here, that's the only thing right and next one.

G

So the purpose is to give guidelines to protocol designers at layer, 2 and tunnel protocol designers on how to ensure he CN is correctly put into their design and Floyd cross-layer, and we've also been doing some liaisons with other groups that designing layer, two protocols and systems with layering in them to catch any systematic layering problems with ecn with on one with the I Triple E that's completed and very easy and 3gb p is ongoing. That's what I'm going to talk about other reason.

A

Well, so the the.

G

Activity on a 3gpp.

A

G

Have a response to.

A

G

Them in December they identify, he asked them to identify working groups and specs that that reference dca they identified six working groups and I'll put the abbreviations there. Yuvika.

A

Three to compete.

G

Speak and twenty spares that reference, ECA I, 31-6, De, Los, Lagos,.

A

G

79, which is ACN in RTP, and so next we break them all all 20, really big space, ok, strictly I, search for LSU.

G

And prepare to admit to the chairs yet, and others me and John kapiti compiling reveal who works in the freaky BP, so I've got another slide on the other recent activity, but lighter from this graph is rated to another legend critical trill presentation earlier this week on that, oh and that's actually redesigning easy and supporting drill, because that changed trill and finally, there's been some discussion on how to apply a large number for both static on space I'm, going to talk about what a couple of slides on the first one one on each on the vast, the 3gpp liaison context.

G

Don't forget to read that. Essentially it says we believe there could be problems here. It looks like you've not may be fully understood, iris lee 3168 and that's about obviously 3168 again, if we've already said is quite unclear and just got bigger and bigger.

G

So the context that way of talking.

A

About is mostly go.

G

But you know the other way station, which is a lie to device and GTP. Tunneled information comes in and it goes out pptp which you pack it take to something convergys protocols and run so anyway. Next slide, these are all the specs I read.

G

Most of them are about yeah and the poles ones are normative, or they mentioned easier. Normatively.

G

Only mine, 3gpp, normative rules are different from ITF and the italic.

A

G

Bold italic ones, sorry, the normative ones were the ones that we were told mentioned: ecn normatively and the italic ones ones. I found some normative easy and texting, but we were told.

A

G

So I it made a difference in my definition of normative to yours, but they might offense insureds other than whether there's normative public it anyway.

G

Top of a summary all nearly all of those are about ecn in RTP UDP and all of that stuff seemed fine. It was people it just made me realize. Just how complicated mobile network is, how many boxes have to say.

G

Yes, we understand ECM before he ain't going to do anything, but the ones that seem to be more problematic for the radio access network ones which the last two and second column, okay, I've got another slide on and also to a certain extent, the third one at the top there 23 401 on GPRS, but otherwise everything else they've fallen apart, see if you find next, so the texting, the random ones you try and Andy evolve. Utah is there I've shown this text last time, but we didn't want to do anything about it.

G

We didn't want to talk to the 3d people about it and this until we found. Maybe there are some other texts that qualified it was something like that, but now I've seen them all I think this is a bit efficient.

G

Two problems with it. First is talking about reading and writing the ECT, no point on a node that isn't I peer to be idea where it's a. If you went back to that, layering diagram I showed earlier it's a layer to these, the under IP node, so I think it needs to be clear or what it means when when a NOP is reading and writing the IP header and secondly, the mark on behavior.

A

Not just the protocol mechanisms.

G

But the behavior is very unclear and I. Think that highlights a difficulty enough having 31-6 gauges matches where as much to blame it like so.

B

Next yeah, so maybe this is the reason why we're trying to write this document in fact yeah yeah yeah. So it's not.

G

B

G

Show the layers me across from the leave the user equipment and I've highlighted here some of the protocols going on so, firstly, there doesn't seem to be any spare anywhere that refers to I've, see 60 40, which says how ACN gets propagated in tunnel and I. It must be tamil hard to have all this GDP stuff going on that that ought to I'm, given even the level of detail at which RTP has been specified for all the middle policies. To get it right.

G

I would imagine also that the tunnel endpoints have to refer to obviously 6044 there a. I hope this largely 6014 isn't clear that it covers.

G

Gtp and that's going to be on the last slide.

G

Secondly, it talks about a PDC p header, which implies that it's only thinking in the downstream direction and downward those oubli also introverted, investing in the app's rating when it would be writing into the GTP italy, NOP and.

G

Then, when it's as easy and co point in the PDC p as to use which are, I think it must be the easy and header inside there, but don't think edct has gotten easier ever you want care took sam, I'm wrong, but I don't think it has, and so you know there's these questions how's. It know they're, particularly because there may be a robust, robust header compression, especially in this is a voice service. So the IP header inside I not have an immediately obvious to bcn built. It may be compressed no yeah.

A

G

I'm not sure how anyone could implement this right.

B

So, at least in theory, some of the rock mods will transport those bit but not be visible because will be rock compressed. Yeah yeah.

G

Okay, so we will put this into a formal yeah because Lecter, that seems to be that seems to be problem fair than a violent. No one can implement this or that they'll be into non interoperable and logistics, presentations of it or whatever, doesn't see enough.

G

Okay, the thing I speak on that before I move on to a trailer and stuff Your.

K

Majesty, meaning speaking as individuals, I'm, not not a few GBP radio expert so but I su bloody Rock compressions just happening on the radio pod from the unit B to the terminal, and that did that you know be, can let's say, he's doing the compression and decompression. So he sees the bit yeah and I think you're right and interpret any saying that the PDC BSU is the userland IP header interests in action which is strangely described but I. Think that's the reading.

M

I'm gear up, Maya I'm, the liaison responsible from sweetie PP towards ITF I'm, also not a radio guy, so I can't help you with that. But I want to say thank you very much for the great work at it and for even opening dead, specs like 23 or whatever. That was so that's really impressive, and thanks a lot and I hope you get this rolling and we can maybe talk offline them. How to best push this into the right working groups thanks a lot excellent.

N

Yeah Kevin Smith eye definer, thanks, Bob, think our greatest great work and we talked about hcn. Could this also scale two things up connects, for example,.

N

Yeah, so could the work we make a bit more generic, so to speak, to say: well, it's not just the ecn, co point that others we should be getting access to.

G

For this drop around so long, not extended, scopus I think we're about to publish it so I'm, one of them right through there yeah sure.

N

Okay, we could take some of the thinking, maybe with in 3gpp yeah thanks.

G

Right there is one more sites are about three PP which about the Mafia well away, but it's that he talks about marking and congestion is, if it's a sort of on our state, and it gives the implications that it will then start setting the sea bit all the time. And then, when the injection finished it was people turning off. It's not absolutely clear, but if the codecs are expecting that and that's what they respond to.

G

If you get some congestion in a normal TV network outside it's going to not doing that and and so the credits could really feel with other or we have to standardize, one which I think you it'd be best to spend those and more randomized behavior that is normally make. um But I don't exactly know what intended or what anyone's implemented. So that that's a bit more difficult issue, because I know the 3gpp probably doesn't want to spend those too much about behavior.

B

I think we've seen these issues talked about before so important things to call this house and also to reply to them and see what the feedback is.

G

It's more about constraints on behavior that doesn't seem to be specified, it might be somewhere and I'm founder, yeah right next trip right trail.

G

The only reason I'm bringing this out is. This could be quite independent, but before we put, the driver I think that the draw a new mural work improve our school, but we had an idea on how to do easy and supporting trail that if we both really need to the drawers as a more general case, because there's a there's a list of ideas, perhaps or ecn in there and the way we're doing trailer is hopefully this way. When we presented this morning, which I said this is really cool.

G

If it's a it's a really easy neat way of doing surveys and if the reason it's cool is because Trillo has been properly designed for forward compatibility for things that come along later, and it's made it really easy to do all right, and it's not yet in that.

A

G

It will be writing into that draft within the next 1.5 months. I believe, is the current fashion.

G

So when it's on 01 see the idea where we can get the slides for the session early right next is updating some specs, and this is a more processed question. We ask this on the list and lay flat homie reply, and so I'm going to keep asking until I get require one.

G

The the problem is that when RFC 6040 was written and I wrote it, but obviously we all wrote it in the end, the it says, but if you it was it scopus, Becky, tunnels and.

G

He didn't cater for the case where you've got an idea, nike tunnel with the shimmy in between H, which can't exist independently, so it has to be put on by the same node here as the IP header, because it's a it's definitely a shame. It's not a header, and all we really needed to do was just extend scope just a little bit to cover all those places, and we did it so I'd like to update as a proposed standard that what we're talking about here is vcp and I.

G

Don't think well, they'd be like I said that any advice is that a bcp cannot later proposed ended with a maybe the brute. Is that correct.

B

Probably, but look less less, not less, not and we're not having that discussion here. So you just tell us what you think and then we're going to discuss. You don't work to look a wall chair. This is another.

G

Another say only proposed standard as a piece for 64 I think would be a lot better by the cleaner so wanting to implement 6040. They don't know how to read this massive long, easy any capsulation justify this one. Sec I think.

B

We need to be very clear about tunnels in general, because there's a growing number of them and we got to offer good advice to people making them and I personally look to stop creating new tunnels to keep doing things, but well. I can't predict the future and then I suspect people will steal still find new uses for new tunnels and all the one who's taught about tunnels is nothing like a little mr. tunnels. Eric I'm Mike.

J

I just heard about the new one, its IP / UDP came up this week, but we're getting better at creating them faster as far as I can tell, but whether that's a good thing or a bad thing, yeah after you um I, think being able to capture sort of a paragraph saying this is the behavior. We would want right and having that be fairly condensed, and then we can figure out how to distribute it, because some other, when I saw the X line on this back right.

J

So that's not a product of the idea and it so that particular RFC is a bit tricky right. In terms of you know, we can have some other document that refers to it, saying if you doing the X client, you please do this and we'll make sure it goes into the current set of envy or three drafts, the expand, GPE, etc. Jodi you right GUI, but but in the current one and in.

G

G

Achieve that thought process, because the machine is putting a PTP in a row, IV editor but I think the way around here us to say when they don't hit or a lot of it or even if it's assuming between this is how you propagate emotion between and then three GBP or the egg zactly refer to that.

J

Firm, yes, and that well I, don't know my bcp versus PRS. But to me that sounds like this is protocol behavior that we could standardize, right and and and where you apply it is a separate matter. But yes, so.

B

Bob, without prejudicing my options, would you be willing to write an internet draft which just catches, this particular bit of tunnel design thing as a separate document, and then we can figure out who should read this and how we should capture it. Maybe some of it you're going to the design team activity in the routing area. Maybe it should be copied, maybe to be reference, maybe should be a separate document. I don't know, but having it as a separate piece of text. To start with would be a really good thing, yeah and then.

J

We could, we could add references to it from other places and make people aware of it, and if it's you know a page to read or whatever it should be easy. Yeah.

G

Most people think we're assuming that our FC 64g applies to them and it just clarifies if it does apply.

B

Okay, next steps.

G

So yeah I will give you a response to the for the liaison and that will go through the approval process. Presume mission I, think I. We, the co-authors, will help you the bash. There's a couple of eating and calves habits to do, and you just asked me to do that and you draw first so pull it there. The second one I'll just have to drill for technique and then I'd like that. If we can get that shot up for a working group last call or something is think it's been around enough.

G

Isn't it we've got a few reviews down a bit.

B

Yeah, yeah and I'm thinking about whether and how we know we got to working group lost coal status when we're trying to inform people in other places about what's going on, I mean if, if we knew that the advice was being taken well and they were understanding it, then that would really feel immediately jump to. Let's go for a working group last call and I'll go. This is a bit of a unidirectional pipe. Sometimes so, and we can offer the advice and we don't actually require a response.

B

So, let's it by bearing that in mind, ask again when you've done the edits and you probably got any questions about the draft. Apart from what we've taken interactively.

B

The intention was always to get this thing published so that other people would start designing things in the way we thought they should be designed as in TS VW g. This is advice to the community on how to use ecn in places where people might get it perhaps wrong if they didn't actually think about it carefully- and my guess is you've captured. Most of that.

B

So please read this draft we'd love to see comments on the list would look to see, notes that you've read it and that will help us figure out whether we're after working group last coal position how, but for the various things you've no promised I'll make notes. You'd have to be able ambitious. Just tell me in real numbers, 1.5 months for everything, it's a new standard.

B

B

Good 1.5 months, second right.

G

B

Went to the session end in four minutes, so you have four minutes and you have a presentation.

G

This is bringing monetary ecca semantics for this much larger agent, with Horace Lola t a loss scalpel to those next you may have seen just so. You know which thing remember to report for the emerging demonstrating prize, very loyal, didn't see it.

G

The idea is to incremental irre plac best areas so that all track and have low low and low loss, and also you can get rid of scalable. It's kind of property, CP, yeah it'll be so much better. It's going to get Floyd is operating length operator to see the value they can build new product. Next, when.

B

We have two minutes: why don't you cut to what you're asking a Wiccan good? You look up.

G

G

G

Point is there a need for three twice standardization, but what about here is the thing to classify also who can get it right here, and this is a type of eca. So what would housing is to use the ECT one coke going to do this next?

G

The appendix of this drawer looks at other possibilities. Threea showing is tight. Yes, p, + sen e CA, donne 0 set1 see her point as well. Surely, I've got trying to explain.

G

The reasoning is all recorded in that appendix have two other possibilities were quickly discounted, and so basically we pick East e1, Percy, there's more green boxes, and next, yes, dude.

A

G

Come on same with the minus, whatever so putting on 68 but on the ice is the root area to drop in sales. Mom with Michael talked about her GPA has to be the same now. What would prefer to us through here, because we want to be very little different by proposing a new you know, but not quite birthday, 3168 proposing a new service and in fact, in 3168 it say we have decided to make say oasis of search issued, making differently such doc. This is that 40 years on, this may go even saying many PCM.

G

Look that good. So what holy is directly a stake in spray bottle bottom, but we will standardize the draw the if a package can be drawn with the square now the square and from the pack, all the older aquariums were designed around TCP square little wonder that square bombing in for the future. We get rid of that square as where it, and it gives you a long-drawn whole system, which is why we're at very low life. So, in order to feel this, it's a no warm rock. Now we write it for previous idea.

G

Didn't get a small.

B

So pop is an extra one. Love tsv WG should look up next slide.

G

Yeah yeah so kinetic says: what way choosing this will disallow preclude one thing years: it would obsolete, be easier nonten. It says why because there's otherwise doing it, but we found now and also there's very assertive in collusion with a potentially so now. This is poor to kick pick something and we'll be discussing what would be the best thing to do to there and we want to avoid that discussion right, I'm. Finally, not.

B

The sex life yeah more next steps. I next, add your next steps. Don't you yeah I'm, not yeah,.

G

We can oh no Anastasia, so I'm gonna have to say hmm thanks at the moment experimental, but they're good cause confusion. So I'd like a baby on the list as to whether, if we proposed standard, um you know when I get review coming to, please bear that you.

B

Can sell they talked about that on the list, but that's probably a bigger than thing to consider than just document it on the list. So I'm talk about it. First, please do next steps. Please.

G

Review there was plenty of a discussion when we first talked about issue on the.

N

Alienware list it's.

G

Been quiet since that we assume that leads a robbery.

B

No, but, and thanks for keeping roughly two tied where we finished on schedule, all right you want to see the last slide is she's. It's for fun,.

B

Then please don't ask comments on the last slide. Your questions are last night. No comments.

A

B

Afraid I have to close tsb WG and at this point and thank you all for coming, we hope to see you again. Please talk about the drafts on the list and apologies from the chairs that we did manage to balance the Aegean the perfect at this time, we'll do a much better job next time and look forward to seeing you all, either in virtual land or at the next to atf meeting.

B

A