Internet Engineering Task Force Interim Meetings, 3 Dec 2020

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: I2NSF WG Interim Meeting, 2020-12-03

Description

I2NSF WG Interim Meeting, 2020-12-03

A

Okay, so we've seen the note well and now we're ready for paul to share his proposal for every charter.

A

B

Hello, hello, folks, yeah, that's it.

A

C

B

So may I share the screen.

C

B

Okay, sure, okay, thank you. Yeah.

B

Okay, you can hear me all right: okay,.

B

So um this is a slide for our itunes set for working group recharging. So I prepared the the chattering text with the diego ropet. Thank you for his help, so this figure shows a self-adaptive, i2 nsf framework to extend our current itunes f.

B

um Basically, you can see uh we introdu uh this new i2rsf component called the i2 nsf analyzer, which means it uh collect a monitoring data from nsf's, okay,.

C

B

As the fiber repeater, etc, so using machine learning um for feedback information with nsf monitoring data, so we're using many kinds of machine learning such as rnn or other yeah, deep learning technology we can apply and we can make some pw information.

B

So we deliver that information through new interface. It is called the application interface, which means that feedback information can be used to update or generate current skills policy rule or regenerate new policy law and that rule update or new generality.

B

The policy will be delivered over an sfa's interface, so we can see this extended the itunes framework providing feedback control loop, so we collect nsf monitoring data and we process the data with the machine learning technology and we deliver pdf information to security, controller controller, deliver new or updated security to appropriate nsfs.

B

So this extension of itunes framework is described in my new id, so only bytes the itunes framework interfaces.

B

Basically, we pretty much completed the restoration interface, consumer phase interface and I said, phase interface, monitoring, interface, okay capability, um the young data model case, my student, patrick, and I prepared all the revision comment from ist. So I think, next early next week we can uh upload new, updated internet draft, along with the revision letter to icg and or the transport area, some art so monitor interface uh case. uh We got young, dr libya, so we are working for revision after capability uh young data model revision, so other cases passed already um working with rascor.

B

um So I believe this is the right time. We introduce new okay interface application interface, so application interface is, you can see between security controller and itonerceptor analyzer? Okay. So this is uh the text for recharging. So I describe this. One is already you guys aware of this description: itunesf, okay, so the basically um itunessev provides security function, vendors users and operators with the standard framework and interfaces for cloud-based security services.

B

Itunes, super i2nsf enables the enforcement of a high-level security policy, which is expressed according to a user. The perspective of target network, this security policy enforcement in i2 nsf is a data delivering approach using the conf young or less compliant, so where skilled followers is constructed based on young.

C

B

So this is a description of our alternative and then I described the uh framework in terms of a component. I turn as a user security, controller net data security functions, nsf and the developers management system.

B

We can say uh dms so and then I will explain uh one by one right, so i2, okay, there is some missing.

B

I okay, so itunes of user specifies uh high level security policy for a target network and the security controller is aware of the capabilities of the attached nsf using them to build the security services certifying the policy expressed by the itunes user on nsf provides a set of specific security capabilities such as firewalling web monitoring, packet inspection and ddos os attack mitigation, applied security policy rules, dms registers, the capabilities of on nsf with the security controller.

B

Okay, and then I explain another of four interfaces, such as the consumer-facing interfaces, nsf phasing interfaces interface and license interface and monitoring interface.

B

So I explained the one by one, so consumer phase interface is used to deliver high-level security policies from itunes user to security. Controller nsf facing interface is used to deliver reliable security policies from security controller to an nsf restoration interface is used to register capabilities of nsf. With with the security controller, the monitoring interface is used to collect monitoring data from on nsf.

B

From now on, um so we explain the new um goal: okay, such as um feedback uh control, okay, feedback, controller, okay, so the goal of itunes f is defined, a set of software interfaces and their models of search interfaces for configuring, maintaining and monitoring nsfs in cloud environment, including nfv, and the ethic deployment for security management automation. This is a new uh keyword, so security manager, automation, means self-addictive security services in an autonomous security system.

B

Itunes app needs to have a feedback control consisting of security policy compilation in on nsf monitoring for on an nsf. There are analysis for nsf monitoring data feedback, delivery and descriptive policy, augmentation or generation. So this is description of a new our core, and then I will explain um the security management automation.

B

The itunes framework requires a new component to collect nsf monitoring data and analyze them, which is called an uh itunes, f analyzer.

B

So also the iterative framework needs a new interface to deliver feedback messages for security, fallacy adjustment from itunes, analyzer to security controller.

B

A proper translation of the friend friend actions onto nsf capabilities require a well undefined model for representing these actions.

B

So this is a security management related extension and next are. I explain some security vulnerability, such as a sprite chain, attacks? Okay, so we are assuming um or developer management uh developers management system providing a clean virtual machine our container.

B

But if such an assumption is broken, our itunes framework can be broken.

B

Itunesf is vulnerable to inside and supply chain attacks, since it trusts nsf capability declarations, as provided by developers, management system dms, assuming that nsf's work appropriately in all circumstances, as well as iterative users, policy declarations and the actions of the security controller, the restoration of nsf capabilities, the declaration of security policy from either the itunes per user or its enforcement by the security controller and the monitoring data from on nsf are assumed to be genuine and non-malicious.

B

Okay, so, and then I described the okay system, breakage scenario, if one of such activities is malicious, the security system based on itunes safe may collapse to prevent this malicious activity from happening in the itunes framework or detect the root of a security attack.

B

All the activities in the itunes framework should be logged in on either or centralized or decentralized, such as a blockchain way.

B

Also, the problem of proponents and the status of the I-2007 component, such as itunes per user security, controller, nsf, dms and itunes, analyzer needed to be verified by remote attestation, leveraging the current result mostly focused on itt environment.

B

Finally, the current young data models for the itonsf interfaces are designed on the basis of nsf's implemented as a virtual machines and therefore they need to be redesigned for the case where iterative components are instantiated by containers in the cloud native nfv architecture.

B

So I described all text uh the chartering so and then I will describe in deliverables okay, so so the single document for on extension, of itunes framework for security management, automation and this document will initially be produced for the reference as a living list to track and record the discussions. The working group may decide to not publish this document as an rfcc.

B

Another uh deliverable is young.

C

B

Model document for alternative application, interface, okay, third, one is a single document for a probability in a user cases in i2 and service security management, automation.

B

First, one is a document for framework for security, follows translation to sport, the mapping between high-level young module and low-level young module. The working group may decide to not publish this document as an rfc. This document will apply the recommendations under discussion in net mode and offset working group on event modeling.

B

Another one is document for remote attestation for itunes, f component, based on the work of a latch working group, the finally, uh a document for itunes f on container deployment. You know cloud native nfv architecture, so this is uh the deliverable list and then this is the milestone. Okay. So, according to my estimation, so march 2020 we adopt an extension of iterative framework for security management.

B

Automation as a working group document and july next year, adopt a young data model for itunes application interface, as working group document same the same time, adopt remote authentication for iterative components as a working world document and next year november, adopt a framework for security. Fallacy translation, as working group document and march 2020, adopt itunesf on on container deployment and, lastly july 2022, adopt a probability and use cases, consider all our work and we deliver the application scenario.

B

Okay. So this is uh my yeah preparation for the shuttering and next slide. I just report um our hackathon project, so maybe during discussion uh may we can use it. I attached appendix okay, okay, thank you.

A

Okay, so before we begin discussion, I'll roman, our ad couldn't be here so I'll relay some notes that he had so three three things uh first is next service would need to be confirmed on the mailing list and we can't reach decisions uh in this entire meeting.

A

Second point is that uh well, there's a thing that there's a really a small group of contributors and a kind of a silent working group outside that small group. So we need the broader support for future work uh to be demonstrated, I mean we need to demonstrate that there's a larger group that is willing to do work, reviews uh whatever um for these new items and additionally, after many years the working group still hasn't finished the core deliverables.

A

Why is it appropriate to take on new work if previously described? Work is not yet done so yeah, that's from roman, so we might want to answer these questions uh before or during the discussion of uh this proposal.

A

B

This isn't meet.

A

Echo, okay, so yeah; okay, maybe.

B

Somebody yeah some speak up and and then I can answer some questions. Okay,.

D

Okay, just a couple of comments on this when it comes to the code deliverables, I do believe that the core deliverables are are available.

D

There is a myth, one missing deliverable, that is the one on the information model, but, as I suggested, I think that the information model can be uh uh was useful as a as an input to the data module. That is, is not an essential deliverable. We can. We can try to reformulate uh some of the aspects in the in these deliverables that have been well.

D

Many of the aspects that were in that deliverable that has not been uh produced where uh were trans translated into into work in the other deliverables and what is missing is precisely part of what we what we have here. This is one thing. The second thing is true: that there is a a small core set of contributors, but well. I think that this is something that could be said of many other groups means I I I don't know. I don't know how to how to coach this.

D

This I mean it's a for sure, it's romance impression and it's difficult to uh uh to uh to argue against it. But if you look at, uh I don't know groups like uh hrs or groups like uh well, I remember being in the radix group in which we were literally four inside the inside the room with one person that was princess, doing everything and the radix was going on. I don't know, but I'm not saying that this is about number.

D

I I believe that probably it's true that the list has not been very active, and probably this is something that we have to to sort out by doing uh being more active in making reviews etc.

D

In summary, when it comes to the code deliverables, I would argue that one of these code, the deliverables during the lifetime of the uh of the group, was transformed into something else in the second case. Well, we we can try to expand this precisely because, when things like, for example, atom, automation or container a container environment, I assume is something that would be of higher interest for the community.

B

Okay, thank you. Yeah there you go yeah so linda.

C

B

C

Can you hear me better.

B

Yeah yeah, I can't hear you okay,.

C

Okay, so um so looking at those um current drafts, we have right. We have quite a few working group drafts, but we are pending for this nsf-facing data model to be revised so that we can call them submit other ones to iesg, because roman really wants to finish one data model first, and he has provided the comments.

C

So I waiting for you guys to revise the drafts, that's the nsf-facing interface and second like, for example, you mentioned this interface between nsf to the analytical tool that information should be covered by the nsf monitoring data model right so yeah. In this example I do see this is a very good picture. Thank you. um So I see this um blue line from nsf to the nsf analyzer that is actually covered by the current charter right, the monitoring interface.

C

So I see one thing missing, which is the application interface, so I personally will feel more comfortable um in the charter discussion to describe what kind of information flowing through this application interface to see that if the current charter actually can support it.

B

Okay, uh linda application interface case, and this is generated by analyzer, so.

C

B

Interface, providing low data information um so such as, let me show you so this hackathon, we implemented the monitoring interface.

B

Like this, okay, so um so this is our yeah itunes, uh user. This is a skill controller, and this is a dms okay and.

E

B

So we implemented this so uh web filter case. We uh using json, uh let's confirm the net comp, maybe next time we're using netconf. So we got, for example, system interface, some in coming outgoing traffic information and other uh log information such as cpuc disk lab disco usage memory, so those kind of system and also traffic information. This is a important some um statistics, those information collected by itunes, uh i297, analyzer and they're. Using machine learning development.

B

We can analyze that this nsf is overloaded by uh some genuine traffic over attacked by some other sprite channel tag or some inside attack so and then that kind of information generated as a piece of information and delivered to security controller. So that's why we need uh requires new interface, so monitoring cases just provide raw data right right. There.

C

Okay, so um so this um monitoring this in this hackathon, that's really good that you have actually implement this. So my question is: are those information, those attributes being incorporated into the nsf monitoring data model.

B

So maybe um maybe sometime currently uh monitoring an interface uh case just focus on raw data information rather than some processed data.

C

B

My preference is we just providing monitoring. uh Currently, a draft name is uh i2nsf nsf monitoring data young model, but we extended this one for interface, which means configure monitored some some data, um some features and.

C

B

Can collect the raw data, so we um okay um add some xml exam first to configure and collect monitoring data. So and then we can say this draft is a monitor interface.

C

B

And then so we I said, the application interface is a new interface, so yeah.

C

Okay, okay, um so regarding to address roman's comments about the current deliverable right. So what's your plan like um um see, we have an sf basic interface. The data model is waiting for real revision and then, after that, we're going to close the consumer phase interface um and then we're going to con like if you want to revise the monitoring data model, that's fine, um the registration interface. So all those work need to be closed at least um submit for iesg.

B

Right, yeah, so, okay, um so you you can see. um So I think you know the capability young model uh suzanne and I working for that one. So patrick my student, a phd student and I uh almost done so early next week we submit and reserve all issues and once it finished also agreed by isg.

C

B

We submit to new okay and that's a facing interface. Okay, so this is actually submitted um linda. I think, okay, so we and then this one is the second one and the consumer paging also done so. We passed the young doctor review and restoration. Also, we passed the young doctor review monitoring case. We have some comment, media mediate and also minor comment, so we reflect and we satisfy young doctor and then we submit we can call score. So I believe all of yeah interface is almost done and so resulting cases still.

B

We have a time, but we need to prepare for next step.

C

Right right, right: okay, okay, so a second thing about getting more people to participate and review. So um have you thought about like what kind of things we can do to make it? This work more visible to larger community.

B

So I think yet currently at least the hallway and the teleponica and actually the academia case uh skk. My university also sung shield university. Also, I think your korea telecom and they have interest in this one- also sk telecom, so korea, korean sum, subs provider. uh I think they have interesting so and european case uh uh diego uh can help and united states case.

B

Maybe linda uh you guys, you can uh help, inter you can invite other right other yeah at the beginning of our itunes, app uh you invited the uh british uh telecom and twitch telecom. So I think our alternative work is almost done and and then I think right now is a good time to invite them, and so we can um advertise our work to them and they can take advantage of our work, because this is open source project hack, also, a project.

C

Right right right right, so I'm just thinking about this small group. We only have 13 people participating today right. Maybe we can.

E

C

Together um a project plan like um different action items to get more people involved, reviewing the work and we can maybe set up a whole- should contact um and put together project plan, and then we can come back to discuss maybe one month after this and to see how we progress.

C

And meanwhile you guys finish- the ongoing drafts get them completed and then we can talk about rechartering. I think that would be.

E

Right, yeah, better.

C

Step to concrete step to move forward to address romans, uh our ages.

B

A

B

Group, in any case, uh the charter.

A

Should be sent to the mailing list uh for well comments from the larger community.

B

Okay, oh so you said again, so you want. You said uh we need to send this contact charger.

A

B

Playset, okay,.

A

To the itunes uh sometime next week I mean when I get the uh the minutes from this meeting and the recording from this meeting up on youtube and okay.

B

Sure everybody can see okay great so I believe yeah linda and the yelp. um My skk team tried to finish all the revision. uh Young data models uh before next march, the meeting itf meeting one one zero so and then we can uh okay, so recharging work. We can uh mainly focus on so so let.

C

B

Yeah finish: young data models; okay,.

C

Okay, any other questions from.

E

uh I I'd like to ask one of uh of regarding roman's comment uh something that tags off the diego's comment. Can you explain what roman's desire is to have an informational model? It's somewhat contrary to uh past um net, confident mod and even the versioning, that's going on in the net confident mod? Why does he want an informational model.

A

I think the reason is that our charter calls for an information model document.

E

Yeah, but when we, when you chartered it when it started the chartering, it was well, this one could be useful and just could be interim in the working group and then go away. So why is he stuck on it?

E

I'm just asking a really blunt question, because when when it was chartered, they were um benoit said well, I'm not really sure you want to publish an information model, it's probably best as an interim document. Okay, that was fine. We took that as don't put too much effort, put your effort into the data models, so why does roman want to go back the other direction? Well,.

A

Because the charter says someone because we never said that we wanted to drop the information model draft. So.

E

Okay, you can tell him that what we were told was the other way around was you can put it on your charter, but unless there's a real push, it isn't going through the isg with joy. So I mean to be really blunt: oh, come on you, we've transitioned ads and.

A

E

A

So next week I'll send them a question to the mailing list and.

E

Then I I stated that to him when he had the objection did he was. Was his comment. I I don't.

E

I don't feel that I feel the effort to go back and try to retrofit an information model to the capabilities and all the rest is sort of make work that we don't have shouldn't have to do unless he can tell me why we're really going back and doing make work if, if he wants to state that he can state a good reason why um you know he may have a really good reason, but I can't figure out what he wants.

E

I don't mean to be um you know. Ben and roman are very intelligent people, but unless they can tell me why I'm going back and retrofitting a model, what value or purpose is, I have a real hard time, revving up the energy to to review and to put that stuff back in it.

E

So that's just my real blunt question. I I'm I'm glad to help. I've got some cycles and I'm glad to help in the next four to five weeks. Read things with paul, so we can push these over the um over to the isg and over the line, but I thought we were sort of stuck on this g. The capability model is key and g roman wants something. Then I think there's not a real lot of energy to go to make work right now I mean.

E

Am I being too blunt your.

C

Voice, linda, hey sue. Thank you very much for for stating this that's very useful. um I thought we already dropped the capability information model.

E

Why do we need? Why do we need any of the information models if we've gone beyond it again, the the original charter from benoit was pretty like well, you can put it on your charter in case it's useful, but if it isn't and you go on to a data model that isn't going to bother us so if he's got a different model than when we started tell us why?

E

If, okay, it's.

C

I mean I mean that's.

E

That's feedback. Why didn't? Why am I doing something? That's make work when we've got a day, the purpose is to get to a data model, see it how it works, get it implemented, at least in open source, throw it out to people and see if it's useful, going back to an info model. Unless there's some useful, he can read the internet drafts, they were there. I I I'm sorry, I'm I.

C

uh Sue um just to clarify right now we don't have any information models anymore. um The capability info model has been um discontinued and the information.

E

C

Need to be captured in the capability data model, if there's anything.

D

C

So that decision was made in october right, but.

D

Well, I started precisely that's why I was asking I mean, but this that's the only code deliverable I can think of among.

C

D

That that's what roman was saying about not delivering core documents.

E

Right, so that's why I was asking.

D

My my return on.

E

My model was, you know I'll. Let me turn my photo back on well like I I'm I'm puzzled, because I thought I felt like linda we'd finished it. What else hasn't been done other than to push these documents over the line into the isg, I'm just asking, so we can help to move things along and help review stuff for paul because he's been taking.

E

You know the editing pen when you're in the middle of a data model only one or two people can really edit effectively. So did I ask the wrong question? um Yo or.

C

My interpretation is roman is asking the the current working group drafts uh like the one he has provided several comments: um okay facing interface data model and.

E

C

uh Revision from the ulcers, and once this one is revised, then we can close uh the other data model. So those all those working graphs need to be closed and submit and finish all the reviews.

E

Okay, so he's just looking for us to push these over the line, that's a highly reasonable attitude by roman. I I got confused by diego and some of joaoff's comments. So I'm glad I was blunt. You know we're not stuck on anything after november after october. We can just push these over the line and send them through and then figure out what the next step is right.

D

Now that's nice, my.

E

Apologies as well.

D

I I totally misunderstood the uh the statement from roman. If he's referring to documents.

E

D

Draw what I said.

E

Okay, then I think we got a plan. I I paul. I I got some cycles because I my teaching load stops at the end of this week and I don't pick up till middle of january, so I'll try to be good to respond within a day or so, and you can push things over the line I mean, I understand, that's reasonable of roman.

E

If, if he's looking for that.

C

So we have how many documents, uh registration, interface or mainly, is rating for the nsa facing data model get that revised. Then we close others, because all all of them are going through the young doctor reviews now right. So the next one is um uh close. The I think the capability data model has been closed right. So we just.

E

C

No man to submit to iesg to review um so roman basically want one finished, and then he will review the next one.

E

A

I was disconnected for a couple of minutes and did we make any decisions about the information? No.

E

No, no, what linda was saying is roman, isn't looking for the informational model, an informational model. If that's true my withdrawal, my comments, I I just was trying to figure out how do we get?

E

I thought we were stuck on the informational model from your comments. If not, we can push toward the data models being reviewed, I'll volunteer to help paul. I was telling him my teaching loads down um over the next four weeks. I can help push okay.

B

E

B

So yeah next early next week um I will submit you yeah devise the draft capability. Young data model with the region.

C

B

And then, let's see the okay direction from ic and the transporter yeah layer, some art, so I think yeah we pretty much addressed so okay and then we can uh moving out move out with the another facing interface, this one okay and then we move to consumer pacing and then next thing is graduation and next one is yeah, monitor interface.

B

So uh so my university, we finish our stop the middle of december and then I focus on all data young mother during a winter break, so korea case by the end of february, um so we have a long uh a winter break. So I focus on two months for our yeah.

C

B

Young mothers, okay, okay,.

C

Also paul um for this recharging work initiative. Can you drive together, like a project plan like how do we get more people involved and.

B

C

Different people different potential people who is responsible for what and then we can come back to discuss that.

B

Okay, you can say how to uh gather more uh people right right.

C

Right participants, yeah get more people interested.

B

Interested um okay, so I will okay try to yeah, collect or gather yeah more people. Okay,.

C

Yeah yeah, I was just thinking. Maybe another aspect, I'm just listening to your talk. um This application interface might be useful um for 3gpp, um like 3gbp, you have 5g core right and then you have the network exposure function with application controller application function. Maybe they need something there. I'm just thinking a lot. Okay,.

B

Cool yeah you can give some or some static, yeah, okay, otherwise, later on, okay, okay, good.

C

Okay, so that's good! Thank you very very much. This is a good meeting.

B

Okay, okay, thank you.

C

Okay, thank you folks. Okay,.

B

Thank you: okay, bye-bye, okay, okay, keep safe; okay, bye-bye yeah, bye, keep saying.

B