 Great, so let's let's get started. So First, please make sure you add yourself to the to the agenda in the meeting notes If you're not able to add yourself then speak up and we'll add you on your behalf Also, is there anything that anyone would like to discuss that is not on the That is not on the meeting notes added a topic on making progress on first responder use case Input coming from the use case discussions Yeah, thank you, Rob. You actually did this exactly the right way Which is what I headed out and something with the agenda. I would encourage those folks to do so as well Okay, so we have it so we have recurring for events we have a few recurring calls we have this one we have The NSM doc use case which occurs every week which occurs every Wednesday at 8 a.m. Pacific time and we have the NSM use case which occurs every second fourth and fifth Monday at 8 a.m. Pacific time as well. We are also participating in the CNCF telecom user group Which occurs every first and third Monday at 8 a.m. Pacific time The birds of a feather for next week has been cancelled due to the fact that most people the most of the people who are driving that the this particular group will be a cube con and There is going to be a talk on this is going to be presented by I believe Dan and Dan cone and and Taylor is that right Taylor? Yes, I'm I'm also sharing the Green there the link to the talk so Yes, sorry. I was muted. Yes This is and that problem title probably needs to just be updated from boff to tug so the intro for the telecom user group will be happening in there as well as You've seen a test bed is going to be related, but it's not it's one initiative related to the tug I'm not sure what all is going to be gone gone on that. That's still getting updated With Cheryl and Dan I'll be mainly focused on the test bed myself. So we also have a CNCF working networking working group that is ran by the CNCF that occurs every two weeks on Tuesday and That occurs at 9 a.m. Pacific time we have next week coming up. We have cube con and The cube con itself is from May 21st through 23rd. We also have a few co-located events Those colloquial co-located events are vital mini summit Cloud Native Network service day and we have talks in both of them as well In the main session and the main cube con session. We have an intro and deep dive in in the maintainer track so If you so if you'd like to learn more if you're if you want to if you want to join in help describe when a work service message to others Feel free to feel free to join in if you will be in in Barcelona We also have cube con China coming up at the end of June in Shanghai We have an intro talk that will be given by me and Nikolai We have on s Europe coming up Which is the call for paper closes on June 16th? So if you would like to give a talk there and If you if you would like some help putting something together Come come reach out to any of us who are To any of the committers and we will help you set up your your talk We have any of 2019 we have and cube con in November Well, both of them unfortunately located at the same time in different cities But one of them is in Los Angeles and cube con is in San Diego The call for papers for cube con is currently open. So again, same thing as on S If you'd like to give a talk there, you know, definitely speak up We will definitely be submitting a few network service mesh talks there as well If you have an event that is not listed Definitely speak up and also open a pull request. There is a link to the site and And so onto the social media community team Lucina you have the the floor Thank you Sure in the last week We gained 13 followers for up to 185 people following and service mesh on Twitter We've followed 200 more. So we're over 1,000 following accounts and We have 21 tweets. That's four over last week and retweeted about 10 things and I scheduled five tweets in Hootsuite to go out Kind of one a day to promote the intro to network service mesh the deep dive as well as the talks during the FTIO mini summit and the LFN session If there's anything that anything else you'd like me to schedule for I'll be happy to do that today and then I May set the account on autopilot until Cubecon, I'm also presenting to talks at Cubecon next Tuesday and Wednesday to prepare for cool Autopilot sounds great. I'm delighted that you have a scheduling tool going. That's awesome I feel a little silly that I didn't realize that was a thing The other thing I do suggest is if you have some other network service mesh related thing For example, I know we have people if you do talks and demos and booths, etc Two things that I would strongly suggest you do one is update the events page on the website and the other would be let Lucina know so that she can schedule things because create me if I'm wrong Lucina Adding at your leisure something to the Hootsuite for scheduled tweeting is not hard asking you to actually pay attention and live tweet things is Yes so so Which Also brings us to the frequently asked questions page. So That's something that I'm going to get started with Today, and I'll shop it around on the NSM in the NSM dev Slack channels that people can add or change their Views on things so that way that that way we have something ready for for Cubecon we have So Prem and I have been working on a on getting a meet-up set up in the San Francisco Bay Area Prem do you want to talk about it? Sure? So thanks for it so the idea of the meet-up is essentially to Start with a network service mesh and then go on to have Related talks example in a sum with NY and aspects like that so in a way Frederick, I was able to secure the funding at least for the I'll connect with the team and then The first one Also, just calling out in case of Anyone else interested, please let us know please Frederick nor Reach out to me and then we can definitely look on how if you want to host it Or if you want to partner in doing the meet-up, that would be good. Yeah, so that's the update and as a next step I'm working with the With our company to get funding for the meet-up.org. It comes with a charge we'll probably predict we can probably create the meet-up page in meet-up.org and then Probably then we can probably tweet about it and then publicize the event once we decide on the agenda nice and also, I had a talk with Dan Cohen yesterday and He So he also said that if we want to do anything in San Francisco City that The Linux Foundation has a space that can be used for events. Awesome. Oh So I told him that we will take him up on that as well But that the first one on this is going to be is is likely going to be held at Lumina, right as long as nothing as long as nothing goes Sure sideways on it. So So that's that's another thing that we that's another resource that we have that we can that we can make use of. Yeah, definitely sounds good. So Okay, so That's the next steps. I think we will get that going in the meet-up.org And then you can share it with the rest of the team Yeah, I think that sounds good and then once we run the first one and we gain a Sense of the size of the community We can we can start to look at like what is the cadence that we want to to run because one of the tricks to attracting a lot of people is to have very predictable set of meet-ups with high quality content and So if you're it says you're predictable people know to block off their calendars at a specific time And it just becomes part of their ritual Yeah All right, so we have my pleasure So we we have Since we have cubecon next week We the question is do we cancel next week's call at cubecon? And I also wanted to take the opportunity since I think we have the leads for the other community Weekly community calls here to give them the opportunity to consider whether they want to cancel their calls or not It's up to them But yeah, definitely this one we need to decide So to help with this the Cube the 8 a.m. Meeting in Pacific time it will be 5 p.m. In Barcelona Just for just for reference I'm super unlikely to be able to make this meeting next week during the cubecon There's just entirely too much stuff going on all the time I don't know how other people feel and in terms of the folks here and who's attending cubecon and who's not Yeah, I feel the I feel the same way and historically at previous events. I've also had trouble with connectivity there Yeah, that's also a thing. I know that 90% of the people who regularly attend the docs call is Gonna be in cubecon so I'm just gonna go ahead and call that one. So next Wednesday. We're gonna postpone docs for sure Cool all right, so Should we go ahead then and say we're you know, we've already got you know the call on docs is called for next week Should we say we're gonna cancel and put probably on the meeting? Site that we're canceling next week for this call as well I think that we should for this specific call and The last one is going to be the the use case call And that's rump. He's just rump. He's in Prim's decision Um Yeah, I think the use case is after that would be on 27th, right? They're probably you're right. Good. I think yeah I apologize. I got the picture on No, it's a it's a little confusing even I'm wrapping my head around these scheduling. So yeah Okay, so let's say Tuesday Tuesday and Wednesday the calls are Let's let's prominently put cancelled on them for this week and then we'll list next week and With that we'll we'll catch up next week and see what's going on. So we have Some preliminary release notes So someone can open it up So we ran this through the docs call as well. So this is This is the initial set of release notes that we're they were looking at So right up front the first thing that we want to do is make it easy for people to work out how to get started so This is where we describe How do you how do you install it? How or how do you link to? Material that describes you how to how to how to install it we want to To make it very easy to find out where it is and link them to to working demos and so Once we've done that and then we describe we jump straight into what is what is network service mesh? So I took a departure from from the norm not that we have norms yet on our own releases but very it's very common if you look at like Kubernetes and and Various other communities that what they'll do is they'll just list all the pull requests saying here's what's changed The problem is that if we do that we'll probably list around 900 commits from day one to today and so rather than do that I just I opted for I Opted so 800 commits, but it'll be it'll probably be 900 by the time we get to it But I I opted for Describing what network service mesh is and what the major set of components are and So there's a couple to do tasks that are that are in there that we need to that we need to fix up but Basically describing this in network service mesh. This is the this is the reference architecture that we're that we're releasing and also make it very clear that the reference architecture is Not it's not all of NSM that it's just a small part of NSM And that the big part is going to come as people continue to build on top of it and integrate their own other things into it Call out that it's a state is now a CNC of Sam Park box project We now have a logo when I wrote when we wrote this we did not have an official logo yet So we now have a logo that we're going to add in here as well And Finally a set of known issues So we will also have known issues And describing Describing that this is an alpha release Don't run it in production Yep, if you if you do run it in production then tell us everything that breaks We Go back to how to get involved So we need to fill in like where to find the meetings where to find score where to find us on slack mailing list, etc, etc and One last part is we have a document that we've been filling out and is basically a list of people who have helped make never service mesh what it is today and so We're asking people in the community to add yourselves to this and It doesn't matter what with how how small the the contribution wise I mean if it's just showing up here and asking questions on a and Well, we've had people who've come in here and asked a single question which has changed the trajectory of some of our of some of our approaches Today we're here to talk about Is that Dennis I think it's Dennis if you can mute yourself or just I don't know if anyone has the ability to use him Cool, all right. Thank you Yeah, so so it doesn't matter like how How large or small the contributions that you have are like if you've helped with use cases or you've helped with With documents or or so on like add yourself to this list because that really drives that work service mesh You know, it's it's it's not just a small handful of us at the at the top That that makes all this happen like it really is a community effort and there's so much of the community is done In order to to get us here And so we want you to we want to try we want to try to call out as many people as As we can is having help in the space. So The link is in the the URL and if you know of people who have helped that are not here Like definitely definitely have them in as well, do we have any questions or comments on the on the release notes? Yeah, I think that's something else we're going to put in known issues is there are a bunch of known issues about limitations in the kernel Some of which we've bounced into so for example, the Linux kernel Appears to have a global limit of 128 Mac addresses And it's on cable and so if you're doing something where you're programming neighbors as part of your connection context I can't make you mean you could go tweak your system to increase that limit But I can't make a default Linux system actually scale It's just it's not a thing it does So you can download the the kernel patch it Reinstall it reboot the system for a user. Well, you could tweak a parameter and reboot the system It's a doable thing. I'm not fair enough. But yeah Yeah, thanks. So That's a good call out We will have the same problem as well not only with with our prequests, but Systems with how many Interfaces you can have on a on a single system as well, which is part of the reason whether we really need to bring in Well, which is not really we bring it. We've already brought them in but to To encourage you use of things of things like shared memory Yeah, I mean, please. Oh, this is just kernel interfaces. My f's You know the limit is how much memory you have on the system The kernel interface is I think the limit is a thousand twenty four in total You know and that shouldn't be bad for most nodes at this stage But that that that limitation is going to become all kinds of intractable as servers go up and not just for network service mesh for frankly everybody Yeah, this might actually Give it enough time. This might actually be one of the one of one really good reason that People in the kernel community may actually, you know, We may be able to convince them to up that number if if it's not too difficult of a process, but Or make it more flexible It's kind of weird that it is Pegged at something that inflexible. I mean, I particularly the art table one blows my mind I mean, I've I've I've seen as you might imagine a lot of our table implementations in my in my day I've never seen one pegged that way Well, so how do you want to list these type of these type of limitations? Like where do we where do we want to store them? I think we'll just put them in known issues because they they are just known issues They're in some sense. They're kernel bugs and I don't think they're gonna bite most people at this point Most people are going to be using us for all three so they won't use the IP neighbors stuff So they won't hit that limit on our table size Right now most nodes are topping out at a hundred and a half nodes So a thousand twenty four interfaces is going to be much less of a big deal for them. Um, but as people sort of start growing up server sizes It's going to be more interesting Okay, so we'll add that to do known issues them It's added as a as a to-do as well for that specific one because that's going to be important So we have CI stabilization Ed you have four Yeah, so I just wanted to recognize those of you who have been doing development work the last week or so probably notice that we've had a lot of Trouble with our CI And it seems to have we sort of view Andre and a bunch of those guys have sort of rooted down this to a small number of Of root causes one of them was that as we've added new things like AWS we haven't always figured out the Figured out the right set of limits on quotas for things and I think those are mostly fixed now So we would hit a place where we couldn't create the clusters In order to do testing because we were hitting limits on quotas on various cloud providers. The second thing that's happened is We weren't cleaning up. We were actually leaking clusters Which is kind of a dramatic thing to leak and again, then you hit the wall eventually with the cloud providers And I think that's been fixed A third thing that happened was apparently some of the dependencies between jobs weren't quite set right So that you know, we would have integration tests that would try to run before Images were actually built properly and then they would naturally fail And then I think the last thing that we got cleared up was we had some places where There was some silent swallowing of Of clusters not being properly created which looked like Test failure because the test went to run and there was no cluster to run on but it's actually cluster creation failure So hopefully that will get enormously better at this point But I did want to sort of acknowledge that that's been a problem and what it really comes down to is Just Standing up stuff like this in the cloud to run reliable CI there there have been some learning Opportunities there I like the living opportunity It's all about tone Yeah, I I personally have found like If you if you want to work out how a system works, how do you Like how does this? How do you start up a project? How do you how do you run a server? What parameter are necessary and so on So if you're looking actually game editor A project how do you how do you run a server what parameter are necessary and so on so if you're looking actually game that experience Like CI is by definition That Like it's you're you're spinning up assist your building a system You're spinning it up you're seeing you're seeing how it works and verifying that it works And so so you have all of the code there that that describes how how these mechanisms work So it's it's actually a really good way to to get involved with With the with the project so if someone wants to if someone's been looking for a for an entry point and to never service mentioned it's having a little bit of difficulty and You have some previous And you have some previous systems experience By all means dive in You know and we'll and we'll help you Are there any action items that we want to call out on this said No, it's just one of those things where I noticed it wasn't on the agenda And I know that if I if I wasn't as involved in sort of getting to the small number of issues I would be super concerned with what I'd seen in the CI in the last week And so I did want to make sure that we brought it out front and center We talked about it. We talked about what was going on and the fact that we think we've actually shaken this out at this point It'll probably take a day or two to actually be sure cool And I think that that we should we should say it clearly that we have three public clouds plus packet which is again kind of public cloud Deployments in in our CI and synchronizing between these and Making sure that all these work The way that we expect them to work at the same time At the same time, it's I mean it just takes some learning opportunities Before before you do CI is actually incredibly ambitious We run patch by patch validation For a very large number of integration tests across the three major public clouds in North America plus vanilla Kubernetes on bare metal every single time And so it's a very ambitious CI and the fact that you know, we're actually doing it is quite awesome. Yeah, and in terms of In terms of best practices in the application world, so they there's this There's this push towards What they call continuous Continuous deployment and continuous delivery. So this one is not continuous deployment, but it is continuous it is pushing towards continuous delivery towards that direction and It's just exactly as you described any any commit you go through runs through a large battery of tests Things are fully automated if something breaks you fix the automation you fix the you fix the bug and you don't And you don't wait until like a nightly test or a weekly test to work out that things are broken because by then You have no idea what broke things Yep so Nikolai you want to talk about the Android release Yeah So I did and I believe a couple of other people who did some grooming around the backlog here and what's in progress We also merged some things And Think that we are I mean Okay, today we were supposed to do the The release but yeah, still we have some issues but still I think that we are in a good in a good shape Relative to where we are. I mean what What's what is the status of The backlog what was it like three three weeks ago, so I think that we have pushed out here So what we spoke last week was that we are going to consider Creating the release branch at some time around today so that we can We can have a stable set of demos there for the cube con So I am fine with that. I mean, I think that's that we are we are the stage where we can we can do that I mean at least regarding the demos. I don't think that we have any issues there and they should be They should be completely fine. I mean if if we branch today, that's one thing on the other on the other hand we also There is this push around consolidating the images that we use in our CI And I am also I don't know how many of you remember but there is another repo code examples where I'm also pushing some things there and I'm also working on Moving or at least replicating the same examples that we have in the main area put there so My question here would be do we think that the examples could be this source of truth where we actually do the The demos from or we should just keep the idea with the branch and continue with it Yeah, the other one I would want to throw out there is I kind of like to see CI stable for a couple of days before we pull the branch I really want to see the branch pulled not only for the demos of cube con but also because that freeze up master again You know for development and there's a lot of good reasons. I want to see the branch pulled But I kind of like to see the CI relatively stable at least on map, you know as it's coming across on master For a day or two before then just to make sure that you're actually stamped out the CI problems And we're not gonna have to try and double commit those CI fixes Does that make sense? Okay. Yeah sounds good Yeah makes sense of course Okay, so then Then let's just quickly quickly go through the through the in progress things here So we have a couple of things around IPv6 Which seem to be blocked more or less On some Some problems that we found with bringing in To the latest VPP release 2.1 So I'm not sure if If this should be a showstopper for us, I mean like should we consider IPv6 for the release or should we not I think that for the time being we can but Maybe we'll see that down the road other than that. I don't see anything that is really particularly outstanding Probably mostly things around the CI and I would agree with you right here that maybe maybe maybe this should be our Our main point here. I mean just make CI stable and then branch and Yeah, quite frankly what I would really love to see is I still would like to see us push to get IPv6 working If we can We have a little bit of time before we pull the branch because they're going to get sky stable So we can keep pushing on v6 and see if that comes together if it comes together it comes together, but you know Having v6 working and I think would be really really a good idea So IPv6 payloads it really depends on the Just fixing or at least figuring out what's what's going on why they started failing when 2.1 Gets in But for the cluster, I think that that it would be I mean our our CI and everything is so complicated already I think IPv6 on top of it before we consider a substantial Refactor or at least some kind of more constantly dating this whole Big YAML file that we do. I know that Andre had some ideas around changing the CI I mean It's actually one of the reasons that I think I'd be really happy when we get the branch pulled Because the kinds of things that that Andre is wanting to do I think are super good and important in terms of simplifying the CI But I'm not sure there are the things to do late stage on Yeah, of course, of course, of course Now, I think your point is that the world we're too far along to add the IPv6 cluster now I Need to say if that turns out to be the case then Getting the branch pulled earlier is also good because that work can push forward on the On master and what was the issue that we needed to go version 2.1 of the agent for the payloads The issue is that tests are failing with 2.1 No, but why are we having to go to v2.1 Is by question and for v6 payloads. Oh, yeah, because we needed to To enable there was a syscatl to Enable IPv6 in the containers which needs to be done. Ah, yeah. Yeah. Yeah. Yeah. Okay. Yeah, I get this For those of you who have not been following closely It turns out that the network namespaces if you're a v4 cluster They are set up to not allow you to have v6 addresses in them by default And so there was a fix that needed to make that not so so that you could have v6 interfaces for payload dropped into pods in a v4 but only So, okay So that's the thing we can sort of figure out the question I would have on the the v2.1 patches Are we seeing are the has that been re-based on the fixes that we've done for the CI so far and or could it be Because there's a lot of weird shit They got fixed Yeah, okay, so the last push was Seven hours ago. Yeah, it might be worth encouraging him to Basically to to re-update for master because I think there are some things that have been fixed recently and try again Sorry guys, you call I Probably with IPv6 we could check Google Cloud since as I've seen it has both IPv4 and IPv6 IP addresses for pods Yeah, that's not not the problem in our packet also has both addresses. So that's not really the issue I mean Regarding the clusters. I think is that you essentially you will need we will need to have a separate deployment for for Kubernetes cluster Additional to whatever we have now. So, okay, ideally we we should just deploy Kubernetes again with IPv6 enabled and then run the test again and this will just explode all the All the CIs test that we have already. I mean it will be Yes At the moment, I'm trying to as pipelines Could be a bit better We'll figure out So here's what I think I'm hearing right which is the the IPv6 kids cluster Stuff in other words running NSM Cluster that may bump out to the next release because the timing the testing NSM with IPv6 payloads Let's go ahead and see if we can get the Get that patch updated to what's currently you know on top of what's currently a master The test fails that I'm seeing there are look like some that we solved with some of the fixes that went into the last year That really cause infrastructure issues. So let's go ahead and see if we can get him to update this to the latest On master and then let's rerun the CI again and see where it stands sound good Yeah, now real quick for our IPv6 proponents on the call and you already made the call to know in the chat Is my question is Making sure we have IPv6 payloads working on Basically if you had to pick between that and having working on an IPv6 cluster, which one do you care more about Can you ask that one more time Ed? I was you were breaking up in my headset. That's okay So we have two things on IPv6 and and I would like to get both of them But priorities are always a thing and so that I find it most productive to ask people to priority or rank things So for the v6 proponents, which do you care more about having v6 payloads running across the network service mesh? Or having the network service mesh running on a kubernetes IPv6 only cluster me IPv6 only cluster Interesting, okay, that is not what I would have predicted Jeffrey do you know anything? This one's tough. So once again The service provider redheaded stepchild. Um, I have like it's like dictated like anything I do must have IPv6 support So I need both if I had to prioritize I would probably pick payloads because for the cluster itself I can cheat and put an IPv6 zip in front of the services and then I'm just do IPv4 local And I mean long term though, I want both but I would probably start with payloads personally So just a little bit of information of changes in kubernetes coming up. There is in the near future. They're aiming for For a summer release. They are they're going to have dual stack IPv4 and IPv6 working so we may be able to Make use of that in this in this scenario as well. So christmas is coming this summer. Got it That's a very I would say it's a very optimistic Assessment because even the design the kubernetes enhancement hasn't been merged for dual stack support. So I Won't see it. I think it will not hit a quote probably next to releases at very best Yeah, my my opinion is It's It's not there until it's there and so until it's been properly merged like don't rely on that information, but In the near future, there is a possibility that that some of this may May be resolved But we should not rely on it So add to your question about the part the workloads or the the cluster So for me my use case is I don't really care if my cluster is on a v4 money like control plane What's important to me is that the switching layer example VPP and the workloads can do v6 as well Okay, so we see it's more the workloads than the Do That would seem to be the payloads then right so if you've got a network service Yeah, you can actually carry v6. So that that's what I expected actually honestly So, okay, that's good to know that I misunderstood your initial statement So that that's good and this is like anything else for the way All of us who've been around the block with IPv6 We don't do things we know that the way that network service mesh has been architected It should work perfectly fine with IPv6 with no problems whatsoever We also know because we've already found a problem with the payloads That will be false until we actually test it and find the little myths So Okay, so it sounds like we're leaning towards payloads being the more important one then Yeah, I'm in the same boat as Daniel Okay, that's actually good because I think payloads is potentially quite a bit more addressable if we get lucky Then the fixes to the CI that went in late this morning should Resolve all or most of the issues with the VPP agent 2.1 and so we have the potential To have this converge fairly quickly on the v6 payloads To be fair though, I want you to fix all my problems ed so Pretty sure that's been your expectation all along, okay, so then it's clear I guess Um, okay, so the conclusion is we're going to go for IPv6 payloads mostly for the time being and We are going to wait a couple days to for the CI to stabilize like wait mean active waiting Actively trying to make it more stable and then we branch before we go to kube-con and we have the demos there from there That they should conclude the currents status of the release We have I think like 10 minutes. So Ed, do you want to quickly bring up the topic of the stickers and then we give the floor to Ramki to do his Yeah, if Ramki is cool with that the stickers are not really so much a topic. It's just a victory lap Okay, so It's too good, this is me just kidding Sorry Obviously the stickers will win so I know that I know who's gonna win Yeah, so basically Presuming that the stickers that I ordered are delivered properly to me literally the day before I get on plane to kube-con We will have 500 of each of these stickers available To spread around, you know, we've got two stickers that we did one is the very sensible You know circular logo with network service mesh IO on it The other one was Everyone knows I've got a problem with QR codes. And so the second one is my problem with QR codes manifesting itself So the so the second one is the sneaky booth one you can go and stick on people's booths Well, I mean the second one actually if you scan the QR code it will actually take you to our website And it takes you to our website in a way that can be tracked by google analytics so we can see what the response is on the sticker So um, but like I said, we'll have about 500 of each. So if you find me at kube-con I'm happy to give them to people to hand out themselves. I'm happy to give people to head out of their talks I'm happy to give them to people to head out of their booths I'm just happy I was trying to play around with the AR VR tool just to see if you can if someone scans it can essentially show the mis-creator Awesome You can probably let ronky get to use cases. Yeah, thank you Actually, thank you. Um, so uh, you can actually just project the open the use case document. I can that oh, yeah, awesome. Thank you. So perfect, so, um so we Having very good discussions as part of the use case call and narrowing down to a specific Near-term use case that you can work on um, so far the first responder use case came out as something Uh, which is uh, essentially top notch and of high interest and an area where nsm can show concrete value um, and what we also did was rather than just Drawing out a top level use case. He broke it down to sub use case or functions Correct. So we have to thank Jeffery for that. Yeah, thanks Jeffery for bringing in that. Yeah, of course. Yeah, I have Jeff and uh, Yeah, Daniel you know It matches so many I mean all the operators for the um discussions um, and um So basically as part of it We said the first sub use case or function would be the mobile client site, right? The next sub use case would be sort of what happens in the mobile network packet core, right? So basically the second and um, and also The third one was mobile network ran, right? So basically these are the use cases the sub use cases. He broke it down to um, and in that what we did an analysis was sort of um, hey um How can we make concrete progress on these sub use cases or functions? uh, what we realized was there is a fantastic Open EPC implementation available. Uh, so this is basically from sprint and intel They are the primary uh contributors. It's called omac It's a full-blown EPC implementation every component exists and notably What is of interest would be a cloud native control plane and data plane? Um, even that is fully desegregated right now. This implementation is 4g and then they're moving to 5g uh, very quickly And specifically in the 4g. What is very nice is they have a 4g s gateway p gateway control plane and data plane are fully fully cloud native um High performance with tpdk s rov all the options are available Uh, right and you can even um separate out the s gateway and p gateway if you just say or you can package them together, right? um And so basically the idea was so far. We've been looking at hey Uh, what are some real vnf's to on board to nsm and this seems to be One spot on right where we can jump in and and it has got multiple network interfaces For example, specifically in the sktp gateway data plane Besides the k8s interface. Uh, there is one interface towards ran, which will process gtp u packets and there is another interface towards the Internet or sgi interface that will uh, you know emit out ip sec tunnels Um, or just a standard vlan to whatever you choose, right? Um So, um Now, uh with this view what we're seeing is uh, it's probably worth You know for the team and for the community to work together on this Uh, you know on driving this specific sub use case um And here the goal is hey, uh, this project is coming from intel. So the questions were asked, right? You know intel is the heavy proponent of multis. So here our message is very clear We're going to complement multis in you know in automating the necklace network service as you clearly see Multis has is a good specification, but doesn't drive automation. So basically You know everything even the ip addresses are all manual, right? So and then It basically gets into a plugin specific network plugin specific exercise. So Where nsm can really help us essentially you know in network automation in the case of Hardware and software with different capabilities. For example, the hardware may have sri ov or smartnik Or maybe may not have any high performance. Uh, you know, uh attributes, right? So, how do we Automate it seamlessly and uh Keep things simple from s cnf perspective, right? Uh, that's what that's where nsm value comes in and Regarding making uh, even more specific progress our thoughts were hey, uh, why not um Work very closely with the telco working group, right? Um, we have Taylor here. So basically, hey, um, They've always been looking for good bnx and this is something which we can help drive closely working with them And in terms of Not just working together but also utilizing the setup right common setup and then Making rapid progress Which could also lead to sort of, you know, the next step around how we can Deploy this on the packet infrastructure. They're taking it to the further next level Where that again there seems to be Bigger interest in tying to other sub use cases such as you know cbrs, right? Basically from the folks from cbrs alliance who are trying to get in touch with like jorah lindhorn Who is a key person there and um that could also sort of help us advance You know this use case and sub use case is the nsm and packet, right? Um, at least this was the thought process which uh came out of uh, you know several Use case meetings and uh, really thanks to the use case stream for you know helping get here to a level of Concrete detail which we have here. And of course, uh, pre must be besides me a key proponent of this use case It's like the thank him Well, the team might be thanks from here. I think awesome team Yeah, I think one other ask here is probably We were discussing yesterday is uh, we would know more details when rubber hits the road So at least if we can start implementing Some of the parts of it and then also one of the ask here is if We have any cnf vendor who can probably provide or try to Have an implementation based on nsm that would essentially help out Because one thing what we are looking at is take open e pc and then As ramki was mentioning Try to get into the nsm and then develop the Use develop the business logic as well as the client needed for the first responder. So that's the activity That we are looking at and uh, I think implementation is going to be the key Even a simple park would essentially provide us A lot of or give us a lot of confidence In showing the world on how nsm looks from the real use case perspective And also one closing thought here Nikolai brought up a very good point on you know, the roadmap, right? There's several items use cases ci So we do think This can be a concrete driver for several other activities We start off with a concrete use case and a specific sub use case with an edge and then Then that could be the driver for Other specific talks around ci or integration and what we are trying to do Okay, we are at the top of the hour. Um, do we Think that we should drop up kind of fret. Yeah, I think so Okay, well the I guess the last part is um, uh, before we finish up, um Ed do do we have any intention on doing some form of an nsm happy hour? I think we should I think we should do some form of an nsm get together one of the things I want to check with today is Typically speaking on a cube kind of ends. They have an area with tables and whiteboards and things on the conference floor The evening event schedule is going to be insanely packed. So doing a happy hour. I think it's likely to be um problematic But what I'd like to do is figure out a place in a time where we could all get together Around whiteboards and as a community sort of brainstorm some of the the things going forward Because there's a lot of cool things going forward that we can and should do as a community That I would love to talk through with everybody. Does that sound reasonable? Yeah, I think that sounds reasonable and so we'll I think we can do two things. So number one for the people who are here Um, hop on to the nsm slack channel because your slack should still work in in Barcelona the second thing is will also announce Times and dates for any of these type of events that were coming up both on slack and on on slack and on With that, I don't think we have anything else and uh prem add that to the next uh to the next session Sure, unless it's time-sensitive and get a hold of us off Asynchronous with that. I want to thank everyone for your time. Have a great Have a great day and be safe on your trip over if you're coming to Barcelona And we will see you either next week or the week after Thank you. Have a good one. Thank you guys. Thank you guys. See At least half of you In a week