 Good morning. Good morning I have Watched the the cube calm telecom playlist There's quite a few topics that's Discussing that not discussed in this meeting. So as wonder is that? All in scope for in this meeting or that There are a lot of like 5g, you know as computing care security all those topics That's seems to be not part of this meeting Yeah, I mean any of those are I guess available to discuss We have kind of a primary maybe goals that we're trying to reach All of it would be the general topic is Exploring telcom and cloud native and then Being able to pull whatever you would be useful out of that. So The main three things are our main couple of things user stories and and use cases Kind of be one area based on those and then the best practices around that Just digging into a specific area If you know like a 5g specific topic or looking at a project And digging in is totally relevant. I think it would be more of After talking about it then what are we going to get from it as a group and how are we going to use that so For instance nephew has been a topic we've looked at in the past and we're interested in Trying to highlight and pull out best practices for the Onboarding of CNS and those are I would call those deployment CICD and life cycle management best practices Their nephew is officially focused on Network function deployments and the management. It's it's really when you look down at it. It's trying to take Best practices on managing applications and the dependencies that are running on kubernetes and how do you Make that easier to automate. So this will be like Operations teams and that sort of thing. It's not really specific to the telecom and networking They're making it relevant there by focusing on those use cases But what what topics did you have in mind? Yeah, it's such a wide topic. Maybe maybe can start with like if we Because there's so many videos. I just basically listen to them one like just walking Is there like if we summarize from all this talks, is there any like major trend theme? From all those talks. That's what current status in telecom and clown it is Major trends Is that what you're asking? What are the major trends? Yeah, what's the trend and the technology trend adoption? Yeah, it's basically another way saying I don't know It's just so many different topics, how do we describe what has been going on in that series of talks That might be something to to do in the group just we could start writing out some of stuff and Summarizing some of the things that we've seen For example, there are a couple of 5g talks which some new projects and terminologies have never heard before Before that, I know there are two projects that's been going on one is the more mature the links foundation I'll cut my my project seems to be a kind of established project another projects, which is my understanding relatively new, but it's kind of more automated is open networking foundations Athena If I pronounce that right, it's another project. So those are what I know when it's related to Telecom private 5g But a 5g project that's been in the in the in the kubcon session is something Yet different. So Yeah, just curious. What is what is the actual landscape? What other options if you if a company want to implement private 5g for example Well, one of the One of the I don't know we said that it's just sort of like how can we say like especially for magma is That magma doesn't follow the 3g pp and standard so Yeah, that's why it's pretty hard to use in combination with I don't know with nephew or other projects like sylva and all these these These other ones Regarding the trends that you mentioned I think it's it's pretty hard to summarize because It has been happened in many different areas so so many so many changes probably Taylor mentioned just the way to to boarding things with with nephew is It's kind of the the latest one, but the other the other pretty interesting project is sylva, which is kind of also a new way to to to try to Probably summarize it the infrastructure or something like that so There in especially in telco day. Um, it was presented the the use case of Using open stack tacker It's a way to to onboard NF so that means CNF and also bnf especially bnf so Yeah, it's pretty hard to to just like Try to to say like this is this is the right way to do things like just just picking one single technologies Yeah, I think the problem for me is I'm still relatively I'm just learning the landscape so not only Hard to decide which is the right option, but not even Even though what is the option like what options are out there and what they are That's the part that's difficult for me to There'll be great if we there is a list of options and what the difference is like you mentioned magma Minds saying magma support 4g and will support 5g eventually So you're saying that it doesn't meet the 3g pp Standard that's new to me. Yeah, so those can be a very useful information if there's such a list Even if there is no comment on you know, I think it's pretty hard to Make a like description on which is the right choice because you know different different people have a different choices And the and the involvement of a project whether it's going to be successful have many other factors other than technology choice So I think that's up to the each individual Organization to decide which standard to adopt But it'll be great to have a at least a list of choices on what they are that will kind of landscape I think that will be really helpful So, so I don't know if you have checked the the list or like the the documentation provided by anuket Anuket also has been Done a great job criminal reference Architecture and especially for the infrastructure like them just in this particular area they have provided very useful information how to Define you your infrastructure No, I'm not sure that one of the talk at kubcon. Is that duck can you share the link? Uh, no, it was not presented Anything during the kubcon, uh, but yeah, definitely some information there Yeah, that'll be great. Yeah, it's just anything it's a telecom is so vast Just different choices It'll be great if there's at least some effort to list some list of choices on what they are And I mentioned I I really don't know magma doesn't need the 3g pp. I thought is according to the description at least I thought it's a middle of 4g 5g Standard therefore it will meet the 3g pp standard as well So a lot of these are I don't like anuket silva. Those are all larger it's Fills a project becomes ambiguous Saying that word, but they're larger Efforts that pull in lots of different technologies and and sub projects Um, I would say silva and And the anuket Are larger umbrella projects Uh, nephio also Is pulling in Sub project work Like they're doing stuff with Uh operator framework and and other things so then You could say we are going to use nephio or you could say oh, here's what nephio is using Let's just go and take those pieces and we're going to build them out ourselves um so Yeah, I in the in the talks in cube con that's where I got the the terminologies like silva and uh, what the other one And and then uh, yeah, then uh Nephio, uh, I forgot what there's part of the talk. Um, are they freedom professional will be are they equivalent to magma equivalent to n and f or if they either And uh, oh, they're overlapping. Oh, are they which one is that included in which one? I think that'll be all useful information I wouldn't say they're the same so magma is the for whatever Whatever you're gonna I don't think it depends on what the name would be but if you could if you maybe say the 5g core pieces How do we build those and part of them are following 3g pp and part of them are doing instead of using protocols from 3g pp They're doing other protocols that they think are going to have advantages But it's trying to build that core and nephio is trying to build tools that help with The deployment of network functions Um onto kubernetes And and I think open sec if I recall right they have some open sec pieces, but How do you deploy them into cloud environments? and and handle that life cycle management of the The deployment and upgrades and stuff like that Uh future I think is around the upgrades and stuff. They're right now. They're more focused on first Bringing it up and make sure the network network and everything and then aniquette and Silva I think I would kind of put those closer together even though silva actually includes aniquette as a at least The reference and maybe the testing as a sub piece, but those are More focused on building out here's a entire platform To run here. Here's how you can build out your platform. That's based on kubernetes and How you could run workloads? so those are Taking in a lot more Then say nephio, which is just the the portion that says okay, you have an environment Let me target it and do deployments there Um And then you have projects That would be sub pieces of all of that where i'm saying like istio and um ebpf Showing up in lots of different projects, but then you could say those have their own strong Communities in and of themselves If you go and look at something like uh Das schiff from deutscher telecom it's a it's a framework for deploying and managing a kubernetes Environment target for specifically for telecom workloads That's then based on the patterns and the get ops uh type of deployment management patterns and that uses Sub projects that you also see in these others like Stuff that you make if you go look at aniquette or sylva or even nephio Well, there's going to be an overlap on some of those like there's some stuff from ebpf and the doc chef now they mainly use it for Filtering and pattern matching of what are the things that i'm about to deploy versus stuff like Acceleration or security benefits and other things that i see use but they have those so again those sub projects Like knowing what people keep using again and again and it shows up Across multiple umbrella projects these larger ones if you're seeing those sub projects show up again and again it's probably a good indication that there's more agreement in the community both cloud native and networking slash telecom that those projects are They're finding them to have enough value and use right now for implementations And I probably if you know if you're looking at implementing versus just using something off the shelf That includes all of it Then knowing why they have those there is a good idea because I do see things that have changed over the last You know like 18 months or so Where some things have been pulled out and that's something else was plugged in Yeah, thank you tater. What what do you have to describe? Well, first of all, is there actually a recording for this call? I cannot find it on youtube Because what you just described in the last few minutes I just Is just overwhelming uh information for me that but very helpful if it's putting the at least the Meeting notes And that will at least give me a place to start to look for and compare the different project you just mentioned Yeah, we haven't had the scene of working group call in a month with all of the holidays and kubecon They are recorded this call is recorded and they do get posted to youtube Sometimes there may be one missing but we can request that those get uploaded as well Thanks, lucina. She's just posted into the chat the link to the Playlist on youtube All right, thank you. Yeah. Yeah, I'll go Relisten what to what what you just talked about. It's a It's a it's a quite a lot of information and uh, there's a good starting point Yeah, I think it would be great if the discussion like fought in on the same thread and Just get a more understanding of You know different choices and how it relates as to basically if I The next kubecon will be able to follow why those books are talking in this thread Yeah, the other thing that I was thinking is about, um Usually for example nephio It's almost one year older like I mean Compared with other projects. This is very young project. So um There is a lot of expectation. There's a lot of people like Wants to to use it because has been facing for similar issues um, but also that means that there is a lot of Opposition especially to to implement it or try to especially in brown fields like someone who has been using x and y technologies try to Ringlement or like try to replace all existing solutions with a new Problems or new technology could be hard to To try like, um, I don't I don't know what is what is your situation in this case if you have like a some Brown field or legacy deployment Or if you are trying to collect something from a scratch like a new greenfield deployment Definitely that that changed a lot and the decision that you you can take it so I'm Just more of a just a really sure interested understanding the the landscape and and technology because i'm a i'm a database guy and just interested in networking And and telecom being really the cutting edge in a way in a in a telecom networking revolution right now Yeah, so I just like to understand what's the what is going on in this world. That's basically And it's just so confusing sometimes Okay, yeah, yeah, even in that case like also it's pretty hard for for um For telecoms if you talk about the the to say to to have a state list CNS or applications it's pretty hard for for for telecoms to to achieve that Particular goal like I didn't know we have discussed discussed it before right Taylor like about the Trying to achieve a state list CNS in in the group or I don't remember I can't recall either yeah, but Definitely definitely I just described victor. There is a lot of Transformation in this particular Industry which is very quite interesting to To take a look the closer look Yeah, yeah, thank you for all the answers. So I think I got my enough answer today at least okay Taylor's I would like to bring a particular topic that I was thinking especially last week when I was working with nephew um the thing is like In in in free fight. I mean nephew is using free fight jc as a as a As an example, how can they use all these? operators and technologies one of the The requirements that g5g has for the deployments and they require a specific Kernel module so All the working nodes has to have that particular kernel kernel module pre-installed or like Have it before any Free fight you see installation So the biggest the biggest question is how like What is the cloud native way to to to have That kernel module pre-installed it like I mean basically there are there are a couple of ways to achieve that So the first one could be like maybe pre-b cake So pre-installed it or create your OS image or the working nodes with all those kernel modules pre-installed it and load it so in that way when you bring up your cluster using compliance so Is is I guess the optimal way, but I don't know that that's a The cloud native way to to achieve that the second one is using configuration tools like ones that you have installed your Cluster like you're covering this cluster you create ansible of chef or whatever configuration tools that you use and manage the installation of your Kernel modules But the the third one which is also quite interesting is Well, probably yeah third one is is using like a like the kubernetes demon set and try to Load your kernel module, which are obviously privileged container and mounted the model in that way I know that there is a for option, which is also probably using like a Like a red hat solutions. I think that they have like a something called It's kind of a plug-in like I remember exactly a name this may keep it at this kernel module or something like that Which is a custom custom resource definition or like using more Generic way to load kernel modules But I would like to to discuss a little bit that that option like like like using Your own methods like like using demon set with combination of A custom privilege container when contains the kernel module so I don't know. I don't know. What do you think like a You think that Have you ever seen any particular case how to manage the kernel modules in a cloud native environment? Yeah, I can't really think of a a project or some set of community standards on How you would do that Yeah, you should Think of ways that you would do it in a cloud native way, but that's just off the top of my head How you would kind of model it using different tools It seems like Let's see There's going to be modules that are going to be common to have on deployments and looking at some of the kubernetes production deployment tools For bootstrapping clusters and then setting up a production environment We may see something there. Some of them may Be less cloud native and just here's how to bootstrap and not worrying about normal system deployment versus making it Like ongoing where you can swap in and out or whatever else we're thinking but we could look into some of those Trying to think of one of the names Yeah, exactly because even even in that particular case where I mean where you carry like a demon said and the driver container so That requires A privilege container So which is something that we have Encourage not to use it Especially in terms like a security vulnerability and all these things so But definitely It's like a possible scenario like One of the major arguments is like especially in the public clouds where you don't have control where you your Working notes, so it seems like That's a preference for from major of the people like try to to use demon sets To provision some Host requirements or like yeah host requirements Instead of like when you have like a A private cloud where you have full control of your working notes and you can like bk Some always images and using On the boy like I don't know. I mean, I just just want to bring a particular topic because I found interesting last week and I was facing especially how to tackle that in a More cloud native way like yeah, so Some of that seems like it would be I don't know if that's the right word but more relevant for Edge deployments So that would be maybe related to private 5g deployments if we're talking for like telcom use cases so private 5g and edge overlap pretty strongly and I think The solutions That people are looking at for edge deployments would probably be relevant with regards to Worker node requirements And being able to set them up with whatever Attributes you want on those so then you can use at The work the workload itself I think is pretty good as far as using practices like affinities and tanks to have your Applications go to the right worker node and then it would be does the worker node have the right set of Host capabilities and everything else so then If you're using just srv is one thing, but if it's srv plus Various kernel modules or whatever it may be to You know, it could be like security or something on different things, but It seems like a lot of that would be more common on edge Where you're going to have more pieces all together And the the area that I could think of where a project Maybe an umbrella project nearly I guess you can think of it would be cluster api so I was hearing around kubicon and over the past like six months more telecom companies the vendors and a csp talking about trying to Either directly leverage or get things working with cluster api You know two or three years together was Wasn't as much support for some of the Stuff that was considered bootstrapping new host And bringing them in a new node That support is pretty strong now It's been there for a lot of the cloud providers getting new nodes ready, but there's been a lot of work on bare metal And I believe that there's work around that for like the edge type of use cases with cluster api and related projects The sub project cube adm would be one of them, but there's other projects that are related to that So that would probably be where to look into When we're talking about how do you get custom kernel modules and other capabilities enabled on the host And then at that point you can have You know, it could be tied into daemon sets that are doing things as one thing, but various sidecars and stuff that are taking advantage and then You can start using all the other normal Workload stuff like those affinities and everything Yeah, even for nephio Yeah, it's almost like I mean, I cannot guarantee anything, but It seems like a cluster api is going to be the way to provision the H cluster like the the regional clusters so um Yeah, I have investigate in that use case and yeah Um, this cluster api provide us the level of abstraction and also the It follows the What is it the name? The key rm Model like they're looking for So Yeah, so so it's going to be quite interesting once that you have the cluster provisioning by um cluster api is there a way to Hook any Post provision methods or things like Those Especially use cases like Urban modules or Any Additional plugins I think the ideas At least whenever I was Looking at it more Um, a couple of years ago more heavily and we're doing the scene of test bed specifically and looking at what were the capabilities of cluster api To allow Expanding and adding to a cluster all the way down to configuring like how do you how do you bring in and configure a network switch to add Maybe a whole set of New host servers and connect the man and and have those type of connections and potentially different networks and everything else So how do you provision those sort of things but doing it in a cloud native way? So those are like bigger Goals and longer topics than some of the earlier ones with bringing in nodes on systems like cloud providers Where you already have apis, but those were there and I know that some of the provisioning was already happening for bare metal With Like the A few of the Like tinkerbell Yeah, tinkerbell and everything else from equinex But It's it's that sort of thing when you get it far enough and it's able to handle it then the ideas leveraging an api for being able to say hey go and do these sort of things and now you could have daemon sets that actually are communicating with an api that's Doing well, it's basically cluster api but not thinking of just the the tolling around it but you have a A cnf that needs to be deployed and I guess you have management daemon sets or whatever it's going to be I'm guessing it would probably be some type of it seems like it would be some type of Service that's between system services and the Non-privileged cns that would maybe proxy requests or something I don't know exactly what it would look like but you wouldn't you'd still want to try to permit the idea of Applications shouldn't have direct access to cluster api but they should be saying hey I want this type of capability can you give it to me And the cluster says oh I don't have any of those working ads right now but we see them in our inventory lists or let me go ask the hardware would be the other thing so when you're looking at something like An easier one would be a cloud provider AWS or Azure or whatever and then you can literally go say we need a machine that has this type of hardware provision it okay at it now we have a set of working ads with that type and now we can actually go back and tell the application okay we're ready or we're going to schedule you on these automatically and I think that that sort of automatic provisioning of new nodes that have the capabilities would be the same sort of thing that you'd want to do with provisioning of nodes and putting kernel modules on So so basically what you have described is particular this this area So when when you are requesting a particular CNF which doesn't meet the existing requirements and that's that's the connection that we are trying to do with the in this case with cluster API requesting the infrastructure and ones that this particular requirement has been made so we can provision the rest of this yet then the NF like I mean whatever right I'm going to mention Tinkerbell that's another area that's interesting because Tinkerbell is basically came from the sponsored by Equinix so which has a pretty big player in the in the in data center business but there are several other options I'm not sure they're actually there are several other data center players they don't seem to have the same hardware provisioning solution but there are things like cluster bare metal cluster from the former open stack is and also some other other hardware provisioning so is there any talk on like not target is the best understanding of this what is optional other than of course what public cloud provider provide and just from bare metal perspective it doesn't think about anything else that's being used by the telecom by the private 5G market like another alternative to Tinkerbell like more cloud native way that was question metal metal 3 yeah metal series I think this is a former open stack what is that's called the name it came from that yeah so cluster API has a metal 3 hoop spray would be something also to consider looking at since it's used by sig testing and some of the pieces in there would point you to projects that are used by others I think it's still used by sig testing for deploying clusters for all the kubernetes builds and extensive 80 testing and everything it does production clusters I mean there's several other pieces and sig testing but they refer to it for doing production clusters I think that the I don't know what organization in the industry do for example for DevOps Google do a lot of survey what is being used what's best practice and lesson learned I don't know for telecom are there any people doing that for telecom the other end during the switch during the cloud native they she's she's come was like doing a presentation which I really I was really surprised they were sharing the technologies and the decisions that they make to have in their own infrastructure it seems like they are like not using a particular technology but basically what I what I what I saw they were like using Jenkins with Ansible and building something in house so probably I'm just guessing most of the calls are like they have created like their own tools for provisioning that that infrastructure so I couldn't say like there is a non-linear technology or like a good winner in that particular area probably some of them are just using portion of the tools or like creating something in house and I mentioned that's that's the telecom and the we've works company they're they're behind are promoting get-offs patterns I mean the I think that there's a whole group that there's projects and stuff all around get-offs but get-offs is older than Kubernetes as far as the patterns from DevOps earlier DevOps days but there's different companies that are adopting a lot of those patterns and do it should telecom would be they have some open source projects up there so that's somewhere to look orange they have a whole open source labs and they have a lot of different projects I don't remember what they said specifically about their platform but I don't know I don't know that there's something equivalent to like the chef which is trying to build a cloud platform that is literally being used by understanding for running some workloads that those should talk on and orange has something I don't know if they have like here's a larger umbrella project which you can kind of look at the chef like that and there's a bunch of sub pieces like their operator related project and several others for the chef so you can look at that one and break it out I'm sure that orange has something based on what their talks have been and discussions I don't know if there's something that's listed out there publicly right now but that could be something we could look at Azure for operators is heavily influenced by AT&T work that was happening a few years ago they were building out a Kubernetes based platform and it was like a v2 or v3 of a platform that was based in the past and it was a mix a hybrid and then a lot of those core people that were on those projects went to Azure and that specifically are Microsoft and the Azure for operators team so anything that's like publicly talked about from Azure I think would be relevant for both platform and deployments and workloads those would be a couple of them let me see Swisscom we had a lot of Swisscom people the and the community gathering I know that they're doing stuff they had a whole open source lab before that went away but they have a new open source telecom area but I'm not sure like what has been made publicly available projects they had a I don't know if it was GitLab or what it was before but a lot of that was available to look at the orange stuff you can go through I just don't know how organized from what I've seen it's kind of what projects are available and I put a link into the and be the SR6 another routing related stuff that orange makes available you can go look at those and they contribute maybe another area would just be looking at who's involved with the different directly putting some people over on some of the other projects the CSPs or have a lot of involvement and Silva and FIO I'm saying somebody from different service providers there and then in the Kubernetes side like seeing who's actually attending CIG or the I can't remember I always get confused if it's a working group or a CIG I think it's a working group or a working group multi so that encompasses it's going to be a higher level from what CNI is that includes CNI it's from what I saw in the past it was a lot more activity from vendors that were involved but there were a few people from service providers but that would give kind of an indication of where those service providers at least a portion of them are putting some efforts and then you look back and see where they are I know that Bell Canada for instance is very involved and interested in FIO and EBPF and then you know the DOS EBPF and get ops but I also saw some interest in like cluster API and for more to extend the provisioning capabilities so well coming back to the meeting we have 8 minutes left so any ideas for like to conclude the meeting right now or like what do you suggest I think we can conclude and coming back we do have best practices that we had drafted and we never really got to a working session with all the other activities of the community gathering but it was pretty successful maybe I think we can look over some of the things that were talked about and either continue with current drafts or dig in some of the other topics maybe the next call by the way next week there will be the open source summit and doing that time there will be a summit so it's going to be a mini version of the one summit for those presented that will be possible for some of them to present here as well just to follow up or anything new update sure if you have any contacts or Victor if you had any ideas or specific ones that you're interested in then feel free to reach out or at least drop a note in to the working document but happy to reach out to some of them myself and see if someone's interested in that other than just coming and talking here it would be nice to have a goal with some of it like the FIO Victor M we had talked about highlighting and looking at those best practices so then if we're coming and talking to the FIO we can say hey what are some best practices those best practices ideally rather than just a technology over here maybe just for the sake of saying kind of a way for them to promote their project in the CNCSF community and then be part of the future land station I'm down for that I think if we're going to do something like a overview of projects or whatever then we probably should promote that which we can do on some of the places like there's a a couple of Twitter accounts for the CNCSF telecom related stuff so we can promote how this group is going to come talk on this date and maybe we'll get a larger number of people that are coming can talk with the other co-chairs and figure something out but that sounds like a good idea to have some of them come in as long as we have have it promoted and then I would like to have like where are we going from there not just a it's not just a one hour break break into that have a couple of projects and then a follow up or how does this benefit the community alright anything else before we end the call no I guess that's it thanks thanks anyone for attending thanks cheers bye bye