 Step away from the podium Okay, don't I'll rush in at once anybody out there live-streaming So if I slip and fall here in this I guess they can hear me as well. I figured I'll put this on It's good. Okay, ready to go. Let's kick it off. All right. So the first thing you might notice here Everybody that showed up is that we changed the title My friend here issue Decided to put a little bit different title on it It's basically the same material. So don't don't don't get too confused about that. I figure I'll make it a little catchy Go ahead and hit the next one So today we're gonna talk a little bit about Some architectural work we've been doing for the last four four and a half years here at Red Hat Started out as a really small project I Was involved from the beginning Issue has been involved with the telco and that the edge we have another talk later on Saturday I believe in the morning that covers some of the edge use cases and what we're gonna cover here today is a little bit of The background and I'll throw a few stories in there about some stuff that happened in the beginning to get where we're at now point you to a few of the resources and then I'm gonna hand it off to Issue and he will take us through the actual specific examples that involve very large telco deployments and customers that have been doing some fun stuff all the stuff you see today You can find online. Nothing's secret. Nothing's been hidden We have a red hat portfolio architecture center where all this stuff is up for the public So feel free so when this all started and and you all can think in app dev or in actual building of buildings architectures are meant to be Really easy to understand very precise. How many people here are developers? They're done development. So you've often been sitting there trying to design something and you're you're putting together the architectural documents and the functional designs and you think this is all perfectly clear So when I hand this off to my team, it's gonna come back perfect, right exactly what I wanted to build I mean come on. We've drawn the lines of dimensions. We've shown you exactly how many inches things are supposed to be How many people have been in houses that have been messed up because somebody didn't follow the design Yeah, it's really silly, but it shouldn't happen. You hit the next one But these designs turn into this and I know this is really extreme and this is really rare This doesn't happen that often but somebody physically did that, you know, I mean that that's just thought that we think that's stunning But it's not and so we stepped back and thought how can we get together here and put something together? That's a little bit more higher level. I'm not interested in going into collecting a bunch of toga f You know certification very complex things where you have to have done a lot of class work and research yourself to even Understand what I've drawn up. We want to do something where we collect People that have done it before so you can also do the same thing so you can build build beautiful things or at least understand How they're putting things together We're all smart enough when we get out into our customers or into our own environments to put the pieces together But where do you put the pieces? How do you tie them together so that they scale? That's kind of the idea so what we did was we decided to take actual customer implementations and go look at them and one of the things you often run into in Organizations when you're trying to share this kind of information is it is proprietary or it's not a An actual public reference and if you're waiting for public references you're gonna wait a long time and They also tend to go away rather quickly because somebody decides to be a public reference They can get it to be a part of a keynote at Red Hat Summit for example And then within six months they decide I don't want to be a public reference anymore And you have to pull that offline So that doesn't have a lot of value to share that with you So what if we anonymize this stuff a little bit? I do the research I dig into anything I could find whether it's public or not and then we share it in a generic fashion So we up level three different research projects to an architecture that we can then design so what's common between those because you just Digesting a single architecture and reproducing that for you doesn't do you a lot of good when you're dealing with different choices You've made in the past and different Like restrictions in your own architectural environments So we thought if we did it at scale and and brought it up to a generic level that would be more interesting, right? So let the smart people figure out how to tie these two services together But they do have to have these in these locations to make it work So what we did is we did the research and we came up a way of doing an architectural diagram Initially started with draw. I owe if anybody knows what that is that became diagram net Basically, that's a SAS diagram tooling in a browser Don't care what you're using anybody can use it and then we put our own Configuration on top of it and then host it so we suck in the base project from them And then we diagram across it with our own icons our own colors our own lines Templates so things lock and click and all that kind of stuff And we kept it very simple very descriptive very easy to understand. I hope that's kind of the idea And that's the feedback we've been looking for through those four years as we've done this We're now up to 20 anything been released since I left 30 plus 30 plus Including the ones in progress When I left last month there was 25 or 26 that were published which means they're completed and they include screensaver They include the diagrams they include enablement presentations So we do a public one and then an internal private one that includes exact links to the research document that I used for Customers or whatever that didn't have public references. So that's the only piece you won't see There's presentation videos on how to how to look at the diagrams of what we're talking about and what these things mean and Everything you find in those presentations is a direct link to the tooling So if you click on them, they will open up in the tooling with the diagram You're looking at so any of our internal solution architects can then adjust the titles or Changed it into something specific for a customer If you want to talk about them and we've also done Publications a lot of it's ended up on D zone or on an able architect one of the red hat Blogs that are out there Kind of walking through the various phases of the stuff we found a really great example was one of the integral first integration ones I did I wasn't really sure whether I was on the right track until I wrote the articles So I stepped back and kind of walked people through it in the first four I published They had 40,000 page views inside of a month. I Think there was by the time I got done there was 240,000 page views on that series That's crazy. So, you know, people are really really desperate for just just a generic look at how to make this stuff work Because it's complex, you know, the stuff we're doing in the cloud in the cloud native world is not easy It's not easy because you're having to tackle all different kinds of components Next one There we go. So this is zooming in a little bit into what we decided to do with the architectural diagrams This was a really a lot of conversations between us and people out in the field that how can I digest this and digest this with people that are architects at customers and I also talked to several of my good friends that are out there working a pretty pretty high level architectural positions in some of these Organizations one is in an airport in the Netherlands He he he made a comment for example Can you put a little bit more details in the logical diagram because my architects aren't smart enough to figure out how To map what you're calling this into our organization. So a little bit of a description would help These kind of feedbacks makes this stuff more Consumable for everybody that that's what you're looking for. I don't want to build stuff that nobody uses I don't think you want to see shit that you can't understand So we're trying to find a way that we can all sort this out So we chose to use the kind of diagrams you're looking at and if you look at these things here You'll see there's different blocks and stuff and different groupings of components. That's the logical one Not very hard to digest very easy to see how you got to put the different components together And this is an example of the use case it then dived down inside and looks at what does it look like when you tie it all together? And maybe it's put into clouds or maybe it's put into data centers or where do I group my services and that kind of stuff And this is very specific to what we found inside of various customer research projects that we dug into and real implementations Then if you want to zoom down and talk about this specific one You can go down into an individual component and then zoom in to something a little bit if you want to next and With that I'll be passing it off to an example of telco stuff and we'll walk you through a few of these with issue So before I kind of give examples for telco in the same format that Eric described Let me ask how many people have telco background here Yeah, perfect. That's what I'm expecting. So I'll just give a brief history of what's going on in the telco world and specifically focused on service provider For self service, right? So we're talking about, you know, Verizon's AT&T's T-Mobile who provides service behind your You know cell phone. So that's why the title what's powering your network And and brief history is, you know, they have very storied history They've been providing telecommunications services not since 90s Since this, you know cell phones came about they've been providing services monetizing these services You know offering you, you know, text messaging and email, etc And their model was, you know, every couple of years they upgrade their network So, you know from 2g to 3g 3g to LTE 4g and now 5g And then they manage these networks, you know throughout this and pretty expensive I mean in 2021 I think Verizon spent about 18 billion dollars a year for the CAPEX for the capital expenditure, right? Overall, I think AT&T is about 20 billion dollars. It's pretty expensive Each year for them to buy these equipment. These are, you know, equipment from Alcatel, Lucent, Ericsson, these big boxes that do line speed transfers and You have, you know, a firewall setting for VPN, really expensive stuff And then even to, you know, manage these networks So their operating expenses are pretty expensive, you know, it's very substantial as well So we are talking about, you know, in 2020, I think about a trillion dollars That's how the network providers spent all over the world So it's almost, you know, 65% of their expenditure is on operating expenses So the CAPEX and OPEX for them is really a big deal And what has happened meanwhile is that, you know, smart phone came About, you know, 15 years ago and a lot of the services that are being offered On these phones, like, you know, streaming services, Netflix or, you know, Instagram or messaging services, WhatsApp, that's where actually all the Or the applications have been monetizing, right? So these companies become trillion dollar companies while the network operators Are still doing a lot of expenses, but they have not seen the return They're just providing of plumbing or dump pipes where other players come in And lay their application on top and get the valued services So network service providers realize this, they need to change something They first of all need to reduce their CAPEX and OPEX They cannot afford to build data centers and then put up equipment That probably will go obsolete or doesn't allow them to offer new services As fast as their, you know, cloud peers do So they realize these changes and they have been transforming, you know, This bit by bit, first it was, you know, moving instead of these big boxes Moving out to the, you know, off the shelf hardware and then doing Virtual machines running on top, each of these functions So virtual network functions, right? And that wasn't fast enough, it still was taking them a long time To manage these, you know, VMs and they cannot deploy it as flexibly So for 5G, a lot of transition where it's going is in the area of, you know, Containerizing these functions and that's where we're going to talk about Some of these examples for 5G where, you know, service providers are Realizing the needs of their functions and they also need to, you know, Use hyperscalers like AWS, Azure, et cetera's infrastructure to run Of these services, they cannot afford to build their data centers Everywhere, runs, you know, hundreds of millions of dollars to put it together Another, you know, tens of millions of dollars to upkeep And then when they have to change, for example, if they have, you know, Customers in Boston and all of a sudden the demand is coming up in, say, New York, they will have to put the same equipment So they realize that they need to be able to move faster They need to be able to move to a shared infrastructure like hyperscalers So that is where a lot of work is going for 5G They want flexibility, they want desegregation of hardware and software And then, you know, automation to be able to scale and move faster So here is an example in the same diagram type that Eric talked about So this is kind of, we call it a very high-level architecture diagram There, we're going to be talking about, you know, three use cases for Telco One is the 5G core, which this architecture is going to talk about Next one is going to be their operational and business support systems And the third one is going to be the radio access network So we'll just skim through these just to get you a flavor of what's happening here You can see that what they're doing is they're trying to use the hyperscalers For infrastructure services, they are using container platform This could be on-premise running on hyperscalers But that's where their 5G core functions are running, right? And we'll look at some of these And then underneath all this, they have a common management and orchestration layer That allows them to do a lot of things that, you know, like telemetry What the status of different, you know, containers or applications is, you know The cluster management, security and policy management And they have really big regulatory requirements So they need to have good policy and security management So that is where they added 5G core And this is kind of a little bit more detailed view of And I know it's kind of hard to see, but the goal here is that They are looking at data coming in from radio access networks Into the user plane function From there, it's going to access and mobility management functions And session management These are, you know, control plane functions Where once the session is established, they need to be able to handle it As the user moves from one cell tower to the other And then you have all the other main functions to sustain a 5G core Then there are, you know, supplementary functions So these are, you know, how would you do network exposure Or you would do, you know, different kind of applications To provide, you know, 5G core services At the bottom, these are infrastructure services So this example would be AWS So you have, you know, their identity management service They are, you know, network firewall and routing All of this is in use That's all reported to the hyperscaler And then on the left, you have, you know, management orchestration layer Where, you know, you have a lot of telemetry and orchestration happening Again, the container platform is a key thing here So Kubernetes is powering this They are, you know, hyperscaler infrastructure or on their own on-premise So it provides them a common platform They can run applications wherever it needs to be If they need to move from Boston to New York to Phoenix They can very easily reconfigure and run the applications Wherever the customers or the demand is I will, I know we are really short on time So I will quickly move to the next example So this is the operations and business support system So what's happening in this area is This is what the network operators use to run their operations So if I'm running a, you know, 5G network How do you run it efficiently, right? Making sure that all the applications are provided in the service To the SLA level that I committed, right? So the calls are not dropping Or business support functions like billing You get your bill at the end of the month Or if you are prepaid, then you are paying for this And every time you make a call It wants to make sure that you actually have enough balance In your account before you can make the call So all of these things are happening behind the scene For Delco providers and OSS, VSS are the systems And they have been, again, black boxes, right? Providing these services If they want to spin up a new service quickly It was very hard for them to do it So they figured they need to modernize it Using the same approach Disaggregate, remove, you know, hardware and software Hardware is running on off-the-shelf, you know, components Software are these different network functions Or in this case business functions Running in virtual machines or in this case containers And you can move them around as needed So it gives them agility Gives them, you know, flexibility To, you know, spin up new services As the new use cases like IoT or Edge That Chris was talking about, come about So it gives them, you know Same capabilities as their cloud peers have This is, you know, view of what the companies Each of this So at the underlying, you see common thing Infrastructure services by hyperscalers On the left, your central management This is a piece that's changed for OSS, VSS So you have, you know, platform components That we talked about, you know We have core functions like fault management Accounting, performance management And then you have value services like If they want to do AIML Kind of modeling to see where the things are failing If there is in particular location There are a lot of calls being dropped They can do some model analysis To see what's going on, right So they can do a lot of these things Using value-added services It's like service assurance, right If I've committed to provide you, you know This kind of bandwidth in this time frame Am I able to meet that? And then finally, we will go to the last one Which is, you know, radio access networks So radio access network is a piece When you, you know, make a phone call from your cell phone This is, you dial the number or whatever, right And it connects to the radio access tower, right So that is the radio access network piece For the network service provider So this is, again, very expensive for them to maintain We are talking about, you know, tens of thousands of towers And equipment at these locations So anything they can do to modernize and reduce the cost Will help them a long way At the same time, 5G They want to offer new services They want to be able to offer you, you know You know, VPN capabilities for selected customers They want to be offered guaranteed bandwidth For selected enterprise customers So they can do all this But they have to quickly split up services, right Same thing, agility and how they are doing it Again, desegregating, you know The radio functions, baseband unit And breaking it up into, you know Three different pieces, you know Radio unit, cell side, distributed unit And this is the reason They have, you know, some kind of central unit And the core data center is where we have the 5G functionality Same idea, we have, you know Common layer of management I will go to the last slide Which talks about, you know, breakdown Of same idea, you know Have shared management cluster services You have, you know, the edge data center Cell sites, and then core data center Running the core capabilities here There are some other components You didn't see before, like image registry Or image repository This is to make sure that if they are disconnected From the central site For some reason, the cell tower and region Can still function, you know In a disconnected mode So a lot of these self-sustained capabilities Are built in And to be really honest You should put a lot of work in these These are really complex environments That this is as busy as it's going to get In our architectures by far And I'm not even a telco expert So a lot of work behind this From our telco architects and team Who actually work with, you know Verizon's and AT&T's To understand what their capabilities are How they are breaking it down And then capturing it into these, you know Diagrams And then there's a lot more, right So there is actually if I And the slides are available for download You will see actually, you know How the interaction is happening With this component, right So there's more detail behind These high-level architecture diagrams So you can, you know, take a look I'm not going to go into this But the key idea here is That our telco service providers Are trying to, you know, survive Or actually thrive for 5G Is by desegregating By leveraging a lot of the cloud-native Capabilities By using the shared infrastructure Offered by hyperscalers And then being able to offer services Of, you know, much faster And much cheaper Than what they have been able to in the past So I'll stop here Okay, and before we go to questions Really quickly, two things I wanted to mention So you can get all the details Behind this and more At the Red Hat Portfolio Architecture Center If you look for that There's a, there should be links At the end of this thing with that in there And I also did earlier this year A talk at DevOps in the UK in London Where I covered the tooling live on stage So how to use it and to draw And to do all that It's all on YouTube You can find all that If you can't find it Just reach out to me on LinkedIn, Twitter Whatever you'll find me And I will happily share a pointer to those things I've also blogged about it So if you Google around a little bit You should be able to find it Are there any questions? And before you ask them Let me give you the mic I can't tell whether you're smiling Or whether you're... Alice When you do present one of these architectures To a customer or a solution architect What's, what does the conversation go like? Great question So traditionally, I think And depends on who you're talking to In some cases, these could be the network architect For them, the conversation is I have these kind of capabilities I have this kind of expertise That I can configure Ericsson equipment And that requires a lot of certification How can I have equivalent capabilities In these container network functions? So there is some learning involved From their perspective So that is their concern Is how do I create same capabilities With my certification rights? So for them, they need to be trained On some of these more cloud native way of doing things Other conversation is more on the IT architects For them, it's great Because I didn't know these black boxes They were there I have my enterprise network That I've been able to roll out And update very quickly And now this allows me to bring in these Network-centric components Into the IT fold And handle it the same way So I think depending on who you're talking to Audience, the conversation is a little bit different But I think they all see That that's the direction they need to go And each one of the architect includes an asset That walks you through what it is To talk to one of these stories So we try to give you a little heads up on that I've heard of some telco providers Wanting to run Kubernetes clusters On the 5G cell towers themselves And I was just wondering if you're familiar with that And what kind of workloads they're trying to run there Yeah, great question So yes, that is true So what they have been trying to do is same thing Extending these capabilities All the way to the radio access towers And for them, they have a little bit niche So they're not like generally in enterprise on a cloud You're running three node clusters Or multiple node clusters For them, they want to be able to run single node clusters Or they just want And that's exactly the capabilities that V is red from redhead perspective V started offering single node open shift Based on these required from telco providers Because they needed these capabilities That nobody else was asking for So they needed to be able to function In an isolated environment And for them, the use case is that I have, for example, if I'm a location close to a stadium There's a game on Sunday I need to be able to serve 100,000 customers For three hours And after that, the demand goes away Till the next game comes up to this day Right, so they like the idea of Kubernetes Being able to offer them these peaks and valleys of demand And they demanded that It's not the standard data center Kubernetes But specific features just tailored for their edge needs So yes, yes, thanks One minute left So I don't really think there's time for any public questions But if you have any questions, you'll find me an issue Walking around here And again, we talk on Saturday Yeah, Eric is talking about Saturday We have another session that's focused on edge use cases This was telco And I think that's a 50 minute session So we'll be able to spend a lot more time On different edge use cases Of what we are seeing with the customers But they are doing on edge See you there