 For joining my name is Jason shepherd the VP of ecosystem at Zedita and also an Ella Fedge board member And today we're going to talk about Ella Fedge, you know as a overall umbrella project within Linux Foundation But then also, you know where we're headed some of the key elements of the overall project What the sub projects are you some of the verticals that we're going after in the like You know and with me are my esteemed panelists Starting with Malini You know with VMware, maybe you do a quick introduction and we'll go through the the panel and then kind of get right into it So, hi everyone, I'm Malini Banderu and I work for VMware. I lead the open source IOT edge efforts and I've been involved with edgex foundry for over two years So that's how I also know Jason And with that I hand off Yeah, Vikram Hi, I'm Vikram. I'm a product manager at Mobilegex. I deal with infrastructure Developments there. I also chair the Linux or Linux foundation edge architecture group And at Daniel Hi everyone, my name is Daniel Latharo. I'm a senior technical program manager at the cto office with osis off and I lead the open source program and I'm also a board member at Ella Fedge All right, cool. I'm in Tom. Hey everybody. I'm Tom Nado. I'm a director at Red Hat where I lead networking and a technical partner engineering and we we do a lot of work on Linux foundation edge and networking projects including Ukraine All right, great. Well, welcome everybody. Obviously we've been working together for a while in Ella Fedge and you know, so for The audience. So if you haven't heard of Ella Fedge, you know, it's an umbrella organization within the Linux foundation very similar to how cloud native compute foundation runs also LF networking Where it's a collection of complimentary projects that that we bring in together as a broader community You know projects run independently, but the goal is to start to harmonize across the different projects over time and You each one serves You know a different area in the stack and while on the surface there may seem like some overlap There's also the whole point is to be welcoming as a community and then kind of work through You know any any potential overlap over time and and then also recognize that there's just different choices out there in the market So, you know, that's just how the how open source works But you know think of Ella Fedge, you know as a collection of projects that help developers fast-track Projects for you know iot and edge solutions. So it's not not just about iot. You know, it's a common misnomer You know edge is such a broad topic we'll talk about that throughout this session, but but also You know, how do I how do I Simplified development, you know stop inventing the middle over and over again We're seeing a lot of that happening right now a reinvention of the same things You know that shared technology investment through open source is so important How do I drive more interoperability across different components? You know through an open model You know, how do I decouple my data from any given back end any given cloud? So I'm in control of my data. You know, that's another key element, you know having a multi cloud strategy that starts with an open edge I think we all think is is very important. So a lot of different elements to it But then you know as I mentioned edge is a a nebulous term So I guess the first thing, you know, there was one thing we did recently Was create a taxonomy for LF edge and build standards joke is how do you sell the standards problem is create one more standard So the Taxonomy, how do we solve the taxonomy problem going to create one more taxonomy that we've been getting a lot of good traction with it because of The way we approached it that vikram had led that effort. Maybe vikram. You could tell us a little bit about The taxonomy and just some of the stuff that you've been doing within LF edge Thanks Jason. Yeah, so this came about that We saw so many Overloaded terms and then market and there are so many verticals as Jason was pointing to but we saw a common pattern of continuum We saw there's maturity in the cloud native technologies. Can we take these and extend it to edge workloads? And then we started looking at these problems in in ways that could It's could solve real problems. So we divided this Different edge boundaries into four distinct boundaries Then we talked about what would be the the best ways to look at how enterprise edge would come about what we call as user edges in there How do these edge boundaries sometimes bleed into telco's edge? What are the architecture transfer service provider edge? How can they capitalize on this trend where they can actually migrate workloads from enterprise to Their own network. Um, we suggested few application deployment models there We also talked about some of the patterns we see And in this continuum, we also used some of the mature cloud native technologies which you can extend in these boundaries We also talked about a bunch of these workloads which active projects in the Linux foundation edge are actively working on to solve some of these key challenges But what we want to see and do in the next version is what we are looking for a feedback from you guys to understand What would you like to hear more? We made a first attempt To make it easier for people to understand Not complicated with 5g standards or 3gpp or whatever telco is it talking about but make it easier for people to understand How they can benefit from this newer trend Which is coming their way um Now with there are multiple ways you can actually give us feedback You can join different slack channels or tools we lfh has One good source to understand our paper is We have a video uploaded for our webinar on the lfh channel is titled as demystifying edge You should take a look at it The resource that white paper itself is available under this four resources column on the lfh.org But you can reach us through lfh.org the best way which we think you can contribute is by joining our TAC architecture working group So once you reach out to our social channels, you will be redirected to the TAC architecture working group if you are interested And work with us to define the new things One of the things which we are excited about is now that we have laid out a structure For the technology underneath and how this continuum can work What can we do with the real applications and try to understand their needs and try to talk about the real application work Which can come alive. So this is has been our thoughts. We would look forward to your fact feedback in your future Back to you Um Vikram, you mentioned you broke it up into like four kinds of edge type of things Could you comment about the typical size or the kind of hardware you have at these four types of edges? Like is it one node? Is it 10 hundred something along those lines to help the audience? Yeah, great question. So it it what we saw is We understand the cloud edge, right? This is like a massive data scientist and then there is service provider edge again Same side of the same similar structure, but maybe a little bit less compute The elasticity in the cloud is probably not there, but then user edge is also very involved I mean if you look at iot workloads, that could be just edge gateways There could be some things that the actual enterprise is running there. I'm all workloads. This might be actually a mini data center Um, so different hardware different footprint. So we dissected these boundaries in terms of various Angles one was how the footprint looks like One is like how much cloud native can we go inside? How can we reuse some of the technologies? So if you look at it, that's how we dissected it But we haven't made hard boundaries around these These four things we call as edge what we often see is let's say 5g can offer Private lds and there's a bleeding in across these boundaries So we're not making something really hard and fixed But we want to make it easier for people to understand that these things can work together, right? And then there is There is a continuum existing across them. So defining these terms clearly expressing them. I think is what the idea was for the picture Now we've been getting a I think a lot of good traction You know across the analyst community and just with folks in general because it's it's rather than a lot of edge definitions The taxonomies we've seen Use very nebulous terms like thin and thick near and far Near and far. I joke. It's like sesame street when Grover's like near and far like and I mean it's it's it's a commonly You know well known thing in the telco world, but it means different things to different people and so um Even the the the word real time is very subjective very different if it's latency critical versus latency sensitive You know, no one no one's gonna, you know die if your netflix shuts off Uh, you know might be inconvenient, but but very very different. Um, so latency critical workloads versus latency sensitive That's one of the considerations that we talk about in the taxonomy Is it in a physically secure space like a data center micro module data center? Whatever or is it not you know distributed out in the wild you have to do different things Is it so constrained me to vikram's point that you can't you do various different hardware abstractions You can't run virtual machines or containers Those are the types of considerations that we used to break up the continuum But then also as vikram said there's not like hard and fast rules You know across the lines that it just you know, I think the way we came about it as a community It helps to create more definitive, you know, absolute technical trade-offs to think about the continuum versus the the fuzzy language Also, do you see don't forget like one of the one of the important things we did was to define what edge is or what it's not There's that yeah, right. I mean, it's just sort of like cloud, you know a couple years ago It's it's everything and that really helped break it down too Which is you know, there are different kinds of edge and that's okay and Like we said, uh, we were talking earlier You know one of the things too that it's It should be obvious that there's there's a bunch of common components that can be reused in just really different form factors as they call them or You know the same thing just slightly reused or slightly used in a slightly different use case but Reused so we're not reinventing Whatever component I mean pick anything I mean because we saw this as as the project sort of started to grow in this in this in the lf edge area And that was one of the debates we had early on, right? Which was how many of these do we need and what we realized is that these aren't the same these they were actually different And they were different enough like what vikram was saying That they actually make sense to stay, you know to keep around and care and feed because they're addressing Actually a different kind of use case Well, that sounds like an excellent segue into talking more about a crane up Projects Yeah, I'll talk a little bit more about what the the goal of the crane a project is Yeah, so originally. Um, yeah, it's a good segue. Um And by the way, uh, vikram, I was going to say if people need more, uh Want to give more feedback about the paper you should give them your phone number and they could just Get But uh, we'll put it up in chat. Put it up in the chat. Um, yeah, the the cranial project itself started off really as the The kernel of the lf edge it became the lf edge, right? And uh, when we started the project You know, look, we didn't really know what we were doing in terms of edge, right? As I was saying edge was all of this stuff, which was not core effectively, right? And it was like your uh, sesame street analogy it's either near or far but um We really needed to get down to the details and break it down and that's so a cranial actually started If you look at the blueprints, uh pattern or blueprints Were the way that we tried to break that down and define the patterns, right in the the contextual situation or the use cases that mattered and If you look now, I mean, that's how the project actually does releases They're basically done around a blueprint each blueprint gets a release its lifecycle managed And you'll notice that there are components that are recycled within blueprints You know, thankfully because we're not, you know, we don't have two things we We've created more like one and a half things and we've addressed two use cases bingo, you know, and So you can see, um, you know the the current state of the art And in fact, it's it's instructional to look through the page that has the blueprints on it You can see this evolution. I'm talking about where we started off with I remember bringing this up here was just sdn enable broadband access Yeah, that's that's neat, but that's actually there's a bunch of sub cases in there, right and um, and you can see like If you go to where we are today We've narrowed this down pretty pretty narrowly. I mean there's like a micro mech use case, for example, which took mech which is A pretty narrow area anyway. I mean if you slice off mech in the edge bucket But now we have micro mech so you can see for example, some folks realize There's some special tuning you need to do to actually use this use case And that's really what we're after and that's been real real goodness in this project And it lets people focus on I like you said this Jason earlier like let's stop reinventing the middle and focus on the stuff on the outside Just to evoke the edge word again Let's focus around the edges not the not the middle Yeah, and then the project is is kind of bouncing between Both a horizontal focus and a vertical focus, right? Yeah, and so that's you know the The horizontal focus is really the breadth of use cases so to speak and the horizontal focus or the vertical focus is more like Which platform or platforms can be recycled? I think Milani you talking about this earlier And Daniel about you know recycling components from existing platforms commercial and open source To address the use case and again the the blueprint tries to define the use case and the implement and and implementation Or the components you can use in your implementation and that's a That's a departure from other projects that we or other ways. We've we've approached this ad in the past like I mentioned op and a v or own app in the past Open daylight these projects really focused on either a singular thing Or too many things and they just it wasn't they didn't have a good way of organizing How the project worked and I think these are really You know back to the blueprint thing. They're also a great way to organize the project And a good way for other people to show up and realize where they can contribute immediately as a developer angle Anyway, so yes, I guess tell us more about No, no before that one you had a question Yeah, uh tom, you know, you mentioned vertical and horizontal and using technologies You know across blueprints at the horizontal layer something like maybe kubernetes or open stack or You know open daylight Is there anything happening in that space where these edge use cases are giving feedback to those other projects to maybe Extend them or redefine them or maybe prune them like Like kubernetes is large and it's typically used in data centers Is lf edge may be giving them feedback to maybe trim something or do something a little differently or provide another capability for edge use cases Yeah, that's a that's a good point Again, indeed don't reinvent things that exist department, you know a crane are their examples of that that we've done already uh early on In the early blueprints, they were stack related and those there were a bunch of additions and Enhancements that had to happen to stack that were pushed back Into the actual open stack project And even other projects which I I can't remember exactly but my team was doing a fast and furious work on that early on and We were pushing there and now most recently into kubernetes And and we say kubernetes, but there's like 500 components in kubernetes right now, but You know and that's really part of the challenge too is like finding The component because it probably is one that exists and then modifying it rather than building a new one And and it's important to modify it too. By the way, I should mention this is one of the mistakes We made an opn of e which was we tried to own the component or make a copy of it and then modify it or whatever And it was effectively a fork of the mainline project which then became a real bunch of fun merging all that stuff back so I think you know examples of that are I I raised this earlier like the kubernetes data store is a good example. There was some debate early on about In in one of the blueprints we were looking at about building their own data store. He said, well, no, let's not do that There's one that comes for free. That's pretty good. Let's use that one And and there there's a variety of these examples And they're and it's good that there are a lot of them It means we're recycling So melanie, I guess you know tell us a bit more about Your work in edgex. I mean obviously near and dear to my heart having worked on that since the beginning and we've worked on it together, of course And one of these days I'll learn how to pronounce your name, but Either way Yeah, tell us about your experience in edgex and some of the similar principles that tom was talking about in terms of not reinventing the wheel So one of the things I totally like about edgex is it says I focus on the edge Let me just get this piece right. Let me not worry about orchestration Let me not worry about management. Let me not worry about anything else. So that's its strength and because that's its Main focus you can leverage it in other systems and deploy it and its focus being the edge It says, how do I connect to southbound devices? Whether they're sensors or actuators? Do I support enough connectivity protocols be it bluetooth or mqtt or http? So that's one of its strengths that it focuses on that connectivity And then another piece is it has a rules engine in there so you can do some data processing at that edge So your latency to respond to you know important critical things You know as jason mentioned earlier is a latency sensitive or latency critical You can address those sort of use cases further. Is there some kind of analytics you want to do at the edge? You know, let's say it's a carriage or it's a ev charging type of edge. How often do these vehicles come? How long do they stay? That's what a processing you can leave at the edge inside to pump all that data up to a cloud using a lot of network bandwidth To build those kind of analytical models of these different endpoints And another very nice thing about edgex foundry apart from the rules engine and the connectivity is It says I can export anything you want in any format you want So it has an app function stk that allows you to customize these different outputs that you want to expose Thank you cool And you know and so then we get questions all the time so again as a project where we're Looking to see how we can converse things over time But also recognize where there's there's an inherent different approaches that you need to take and We get asked all the time is it will fledge and edgex. What's the difference and You know I look at it. I mean, you know, I helped get edgex started. So obviously, you know, I You know been deeply involved there But at the same time I see there's a lot of value in the way the the fledge community is approaching things and You know, I think based on the fact that there's a continuum. There's different choices that people have to make Uh, would love to see you know convergence on some of the connectors over time and things like that But but that's part of a whole value of of a community Working together and figuring out what makes sense. But Daniel, can you tell us a little about, you know How fledges approaching things and you know your experience working with that community from osi? Yeah, absolutely. So um Fledge is uh, specializing in uh in the continuum Describe in uh in the paper as a lower footprint than other frameworks, but it also Is position itself in the industrial vertical acknowledging the differences between operations or critical operations versus the rest of the world and um Uh There is a lot of data to be collected. Uh, you know, there's every company right now is going under digital transformation COVID is fast forwarding that and companies have realized that they need to automate and collect A lot more data than they were collecting before and operations has all of these um Kind of type processes and Fledge is trying to help, you know, collect all these diverse data from all these these diverse protocols tiny sensors and combine it to operational data And you know, provide this sort of swift swiss army knife, you know for the edge the operations edge um help fast track development for tiny, you know sensors again different protocols and and and basically Stunderize and curate that data and integrate uh with cloud and other historians and and Products that are used in in those markets or those verticals Yeah, and I you know, I think that uh, you know, I mean the closer you get to the physical world the more, um constrained everything gets the more Custom the form factors get the more Diversity there is and whatnot and that's where I think why I think what we were seeing a number of projects I mean, obviously we have, um, you know SDO in terms of the new project within the community secure device onboarding We've we've got open horizon and we've also got fatal. We've got other projects and and everyone's kind of approaching things in a A little bit of a different way, but we're also, you know seeking to go find that convergence over time but, um You know eve, I'll talk a little bit about you know eve. So so is that either we'd contributed eve? uh You know as one of the the the founding projects and you know think of eve is is focused on Creating a universal abstraction layer for a stuff that's outside of the data center So, you know, a lot of great tools, you know for server class infrastructure and data centers You know tend to be physically secure Got a really good connection between a controller and that data center and You know eve is is a bare metal compute foundation Of course completely open open source vendor neutral apis That enables you to abstract virtual machines and containers Whatever workloads you want from that underlying hardware, which as you get into the day out of the data center world out into the wild It's it's super diverse between x86 and arm hardware You need to have a very strong zero trust security model because people can Access that hardware often you might not really know that you're on a Network behind a network firewall. You don't even know what the network is So that's been the focus of eve So think of you know in the paper we define the smart device edge is is basically any device that's outside of a data center But still capable of running apps and it's the lower limits about 256 megs of memory and then below that you go super constrained Embedded software more custom over the air tools because that because of those constraints The smart device edge includes as we defined it in the taxonomy mobile devices You know clients pcs those are solved problems with ios and android windows ecosystem, etc When you get into the iot sensor world, it's like wild west And and so eve is you think of eve is like doing for iot workloads of the smart device edge what android did for mobile the same component, you know at that edge and so You know and an edge axe or a fledge would ride on top We're actually building certain to build some blueprints and in in a cranial of how do you use those frameworks with with an eve and start to kind of create some of those You know Well the blueprints across the continuum make it make it simplified but So so that's kind of where we're approaching it from an eve standpoint But then of course the collection of projects, I mean Where do we see things you know headed in you know longer term, you know We're seeking paper on the our feedback on the paper You know, we're we're increasingly trying to get more feedback from end users within the community We're starting to kind of spin up that event, but Um, you know there are many any comments on kind of where we're where we're headed Yeah, I mean the the world is exciting now right with at least we started talking with each other and then We are working across the groups So there are a few thoughts we have in mind we can pursue things what people really want to Get out of us. Is there a real application workload need whether you want to run it on? Let's say it's an iod workload on premise or whether you want to run it on network edge What would be what would you like to see? from us Is there some specific apis you're looking for what verticals you want us to address more? What do you think maybe makes sense in terms of business? So There are a wide variety of things we are looking at there are actually a lot of use of cases We understand the workloads now we want to get in a little bit deeper And that's the feedback we really want to see from the market from the actual users What would you like to hear from us more? Yeah, that's that's the idea jason yeah Daniel any any Last thoughts as we kind of wrap up. It's always hard to find the the mute button. Yes. I Sorry my window often disappears on Yeah, um, yeah, I just wanted to mention that uh, one of the things that I I think is very important that it's happening at LFH group is um, the interoperation and interoperability between the project. So Fletch has been working with eve and as well as Uh, I don't know on a blueprint But it's there's some work on going to kind of interoperate and connect between ejects and fletch As well as integrate with us with the open horizon So I think also combining the the power of these frameworks acknowledging heterogeneity and the advantages of that and basically Promoting that interoperability. I think it's key Moving forward. Yeah, that definitely makes sense. So, uh, uh, malaney like final final thoughts I echo what daniel's saying. We're seeing ejects working with e We're seeing ejects working with open horizon for the management aspect monitoring and management And I think that was the vision of LF edge to bring all these projects together have them talk and work together So we have that common framework available for real use cases Yeah, cool. I mean, yeah, I mean even as zanya mentioned, you know Fletch and ejects communities are talking about How can there be a bridge and and whatnot? Um, and tom like a couple got a probably about a minute left any final thoughts Yeah, just to add to the integration aspect. Um, I I'd like to see more, um, encouragement of operators And I don't mean necessarily telco operators, but we're talking about, uh, you know, daniel was talking about industrial edge iot edge We need to make sure we have the people here working on at least contributing to the blueprints that are going to use this stuff operations is the the gift that keeps on giving and Making that as Good and cheap and inexpensive as possible is is really a big deal. So I that's kind of one thing I'd like to see is is us work harder getting Operators, uh to the to the table Yeah, I mean, I think it's key just You know these days if it's a business if you don't have some comprehension of open source software in your business model I think it's very difficult to to excel because you know, you're just trying to reinvent everything You're you're spending a lot of time is what I heard this term a while back and I I love it It's open sources to avoid undifferentiated heavy lifting And there's a lot of undifferentiated heavy lifting happening out there and then so yeah So I mean to enclosing it You're really welcome people to get involved. You know go to the LF bed site You've learned about any of the projects learn about the overall umbrella Always the best way to In part change and open sources to to vote with your keyboard, you know, right code You know help on the next white paper by contributing, you know thoughts there So with that it you we might have time for you a question or two I know we're kind of pushing the overall time but Thank you for for listening and you can find us in the out in the community and and You know have a good one Thank you everybody. Thank you Hello everyone We're right at time and we will probably not have time for the live Q&A But please continue the conversation on the slack channel that was broadcast out It is a slack channel number two if edge project discussions Once again, that slack channel is number two if edge project discussions And thank you for attending today. Have a good day