 All right. Good morning, everyone. I guess it's as good as any time to start. So let's get started Certainly welcome everyone this morning. No last evening was very fun at StackCity probably too fun. So Now let's start talking about telco cloud requirements Before we really get into the specifics of VNF's and what they need I want to make sure we have level set on the basics. So there's no confusion As is always with that telcos, there's acronyms galore. So I apologize ahead of time VNF's and via the word the phrase VNF particular virtual network function, you know Whoever came up with network function virtualization didn't really think through that very well it's now we have to have a reverse of it, but anyways the Today we're gonna talk about we're gonna introduce some of the We're gonna go through the telco cloud and what it is and what we're trying to achieve And then we're gonna talk specifically about the requirements that we're seeing and then possibly some and get into some of the future things that we anticipate As we get into this I want to level set on What we're up to you? It's very easy to get lost as we go forward and go through the this process to get You know very focused on cost and it's solely on cost So, you know, certainly when we talk about commodity boxes and running routers virtual routers on open source and Making the transition to virtualization Most of what you think of is all about okay They're trying to drive the cost down as we heard on Monday from sore up our Network traffic is going up 150,000 percent. It's gonna keep going up and up and we have to respond and we have to respond in a And as most optimal capital spend as possible. So that's certainly a big part of it But we can't be lost on that and only on that certainly there's the operational aspect and and Making things more automated and such but it has to be more than that. Otherwise. It's just a commodity It's just we're just making another making bits into corn We have to add more value We have to find ways to to add more to what we're doing and this is where the cloud is the essential part essential ingredient to that adding value So over time, you know, we've we've expanded and we've done a pretty good job this way And I think you see that represented it most significantly in the in the last year with our acquisition of direct TV That that move has helped us in an enormous ways Beyond just adding more content and adding more services and more of a footprint That's all good but it's also added a group of people that have been very innovative when it came to The presentation layer the content management the content agreements and there are aspects behind the scenes that are are unique in The next period of time We're going to take advantage of this cloud that we're working on now To make it into something make the next generation Direct TV Uverse combination into something really unique So add more value to it and we're going to take advantage of the ability to spin up new new services and new instances of of the content Management system and the way that we work. We're going to do that far faster than we ever did before So that's one part of what we're talking about that I don't want to get lost on and then when we think about the whole process that we're working on You know and we've talked about this in the ITIL world forever that front end of service definition and making that much faster Is as much a part of automation as the rest of it? So that's what I wanted to to make sure we level set on first I mean just to add to a Toby said I mean personally in my view The nirvana will ultimately be when Amazon starts using these clouds to spin up workloads Ultimately because there is a tremendous amount of value in the assets and and and physical Distribution that a lot of these telcos already have you know I mean so there's there's a tremendous amount of value that's underestimated. I think in optimizing that last mile connectivity For a lot of these these over-the-top players and one thing I would add it's more than just infrastructure optimization and for telcos to add New value added services. I think one of the key components is really opening that environment up for anyone to spin up new services Because they're probably going to be a lot more agile and have the ability to spin up creative services and enhance the value To their customers as a result. Oh Don't talk to that. So if you kind of look at I mean what telcos do they're they're Predominantly have done connectivity for the last hundred years in different forms IP In a lot of ways has been commoditized and that's just a reflection of Building an industry that pretty much has transformed the world. So really the next step is how do I start? thinking beyond just IP services or connectivity and really leveraging what I call the beachfront properties that That a lot of these telcos have so if you just look at some of the key properties of what a telco cloud target architecture is and there's Definitely a lot of different components to it, but I'm just focusing at this point on connectivity is physical distribution I mean, that's a given These this is this is the things the web services guys are just dying to have right if you just look at what they're doing Physically, they're they're trying to expand their networks and their data centers closer and closer to as much eyeballs as possible So that's one key component of a telco cloud and of course that enables the workloads to have low low latency high availability through distribution That's natural as a result High volume of last mile throughput that's just a nice way of saying if Netflix wanted to start offering me You know house of cards with 4k streaming and I'm willing to pay for that Netflix can actually optimize that without having to grow their parent points of the AT&T's networks and AT&T doesn't have to Carry all that traffic in their network. I call it almost like CDN on demand or something when I'm not a marketing guy But that's kind of the intent you can optimize your network not for peak but While enhancing the experience so and then I think the last property of course is Telcos have built a very large network I mean they've billions and billions of dollars in the last decade to two decades really building out this massive infrastructure So you can't get lost in the fact that there's this existing environment You have to make the transition for the telcos a lot seamless or a lot easier to into the cloud world So leveraging the existing technologies that they already have deployed and operationalized around and not only makes it an easy Technology shift, but it actually helps a lot and I've seen it in a lot of different Telcos the discussion a lot easier for a lot of these traditional networking and telco guys to start thinking about cloud because There's a comfort level to it. So yeah, so I'm just to add to that point as a bit You know one of the the problems that Marco and I spend a lot of time working on is this Taking a packet and bringing it from the VM that we set up or the container and bringing it to our user Or spring it from one VM to another that that pathway out of from the VM out of the building To another building that has to be somewhat optimized. We can't be Encapsulating it one way here adding another layer of encapsulation there It we have to think about the whole and and to make this work And this is one of the reasons why the work that we're doing is I think is helping is is taking that and solving for latency and jitter issues by Reducing the overhead across the whole the whole of data center and when So before people start freaking out who the CBGP and MPLS I mean this probably makes a lot of sense for networking telco guys I think at the end of the day what I'm trying to capture here is the building blocks that have been used to build networks for the last 20 years or so Looking at building blocks and how large data centers and fabrics have evolved over time if you just build that out You know, you have the data center fabrics All all these combined ultimately a lot of complexity involved a lot of details to build Really large and distributed networks, which in a lot of ways that's what the internet is It's a large distributed network Therefore, how do you kind of bring that into the cloud environment particularly open stack? And being able to abstract a lot of this ultimately to just common APIs and in this case Neutron is kind of the key value of how you bring open stack into this whole telco cloud environment and kind of you know merging the two worlds together and You can build it. Yeah, so It's kind of the idea Open stack everywhere. Yes, that's it took me about two minutes to build all that out Got to be a better way to do that so so yeah, the idea ultimately is you have this physical network And at the end of the day you just work with workload placement placement whether it's containers VMs or even bare metal. I think bare metal or hosting as a service is probably a very Attractive option for a lot of these over-the-top players as well. So the point is Enhanced the value by allowing these cloud workloads to be spun up close to your eyeballs and one key thing for the telcos to realize is That actually gives an incentive for people to want to stay on their network, right? If me as a gamer can pay an extra dollar a month to some gaming company because it reduces my latency by five seconds So I have a better chance of fragging someone while I'm playing online I'm willing to stay on AT&T's network while at the same time playing this gaming paying this gaming company an extra dollar a month Right and they can monetize that as well and that goes back to the whole new value revenue type stream thing yeah, so when we talk about vnfs and The vnfs when we tell you network function these are this is a pretty good Representation of what when I think of vnfs. That's this is what I'm talking about So it's more than just the L3 L to L7 stack of routers firewalls Natting those that's definitely a part of it But it has to be more than that. It's certainly all the security aspects that come with well beyond firewalls and IPS It has to be much more than that And it in the telco world voices a big part It's a lot of our focus But the reality of voice is that it is just one of a billion applications at this time So we want to make it very good But then we want to move on and find ways the bundle that are work with putting it into other contexts and other services as you as you can imagine having closeness to a user in the old web paradigm where the data came from some center and is Distributed out in the in the TV model where we take content from one place and move it everywhere Caching is an essential ingredient But we also have to realize that with iot that the whole that whole dynamic is flipping itself around But now the data is coming from everywhere And if you just think of one simple example like a connected car Having dash cams front and back with 4k video, which is not out of the question Especially the For insurance folks they want to see that kind of Information that amount of data coming into the cloud and being at the edge Needs to be stored needs to be processed So something's gonna have to happen there to change that dynamic So that's that's an example of a vnf that is is is changing the way that we're thinking about And another reason why we're putting these open stacks everywhere The last few bits The last mile as mark was saying the beach fund property that we have We're trying to innovate there and make it much easier and lower cost to bring fiber to your house And so a lot of the work you see us do with cord and With onos around our like volti and our next generation g-pond work is all about that And then the last bit as I mentioned as we Evolve our content management direct TV universe platforms That's gonna be an integral part of it moving and taking that movie that you're watching and putting it into a form that any Device can see I mean just based on that you know what we know right Based on workloads today, right it's akin to web applications back-end databases applications PCRF packet for manipulation You know it's pretty obvious for a vnf in some form or another we got a manipulated packet Combination of traditional networking elements, whether it's sbc load balancing sip UA's or not UA's Back-to-back user agents things like CDN proxies and transformation elements are some of kind of what we know from a vnf Requirement point of view, but if you build us out next what you'll see is we're trying to categorize common cloud and vnf requirements what we see based on The requirements for a lot of these different vnf's which do overlap with traditional cloud native applications and also vnf's that are coming from the telco world, but we do see obviously a very unique set of requirements coming from the vnf's as well And I don't know if you want to talk to specifics there Toby, but yeah, so the key part To telcos want to believe that we're special But in reality many of the requirements that we're seeing from the vnfs I described are very similar to what we've had to deal with in the IT world and I come from a hosting background of host outsourced applications are delivered over the internet and in that realm when you talk about SLA's With a financial company or a government company. There's times when we've written contracts that have hundreds of SLA's in it When you look at the latency requirements of a financial company or the jitter requirements of gaming as as Marco was getting at That's pretty Pretty significant and close to what what is needed in the telco realm So this is an important point that we're trying to make is that a lot of the problems and the requirements Are very similar to ones that have been solved in the cloud world already And where and then that helps us to clarify where where we have actually unique needs So, you know what when it comes to the common parts of it the security and the resiliency and No one wants to see anything go down. There's no no one that doesn't want a hundred percent availability when it comes to Being able to automate everything and our orchestration needs are not something unique There's not something special to a telco's orchestration or monitoring the world of the IT SMs and itils It is very similar almost exactly the same as the F caps world in the OSS world There is almost no difference. So that's that's one of the things that I I Think that OpenStack community is very important about In helping is to solve for this common set of requirements now where we have uniqueness It's certainly when it comes to moving lots of packets and trying to do that in a in a way That's stable and resilient I mean, yeah, I mean I was gonna say those seem obvious at least to me because I've been living this for three years But you know, they're they're just certain things that make BNF's somewhat unique compared to traditional cloud native workloads and that's mainly given the history of the VNF's themselves, right? They've just always lived as as functions in the network PNF's physical network functions So a lot of that is is a requirement as you they transition to the virtual world But as you'll see late as you'll see later I believe that's more of facilitation of enabling the brownfield migration for the very reluctant PNF vendors out there And we'll get into that more detail and we'll get into that in more detail So as we go through this these are some of the high-level methods that we're using to solve for these unique requirements and the one that's top of mind is Getting more packets per second And lower latency and lower jitter and more flows through a x86 server And that's what I like to say is to to SRV or not to SRV I vote not But I think it's a it's a necessity at this point and then Working with the PNF vendors the folks that have made them and getting them to learn more about cloud native mechanisms, especially about horizontal scaling I think You've heard the cat the pets versus cattle analogy the one I like to call pets versus midget cattle Driving toward cloud native is more than that It's a lot more about also about automation and configuration management and and an acceptance of maybe even not having a command line To log into your router These are examples of the kind of change that cloud native is brought to this to the general space Also when it comes to what we're working on We don't live in a greenfield world We have a lot of the old stuff and the legacy and so we have to be mindful of that as we bring things forward and continue to support mechanisms that help to Support things that aren't cloud native so think mechanisms like live migration or remotion These are things that people have used as crutches in the past to help get over the lack of scaling And so that's something that we have to continue to support going forward And then thankfully we can use it for other things like bin packing And then the last bit is I don't think we've done a good enough job at really communicating The especially around networking what we want and what we want to do That's something that we've talked about a lot this week is just the onus is on us to make it much clearer What we're asking for So I'll talk to this particular slide, you know, we've been we've been spending a lot of effort and testing on Trying to figure out for really underneath this There's a lot of each of these lines represents a different V&F and their specific needs And we spent a lot of time working especially Marko And I have been working on trying to test and be able to achieve this level of Performance through our systems so and we've learned a lot and then what I'm hoping to do in the near future Is to create some level of standardization on how to test and benchmark because when it comes to this type of Packets per second our flows through a box There's a lot of variables involved and trying to standardize that and normalize that make it something That's easy to for the vendors and for us to test in a consistent way so we can compare the answers Easily that's that's a been a lot harder than than you can imagine so I'll speak to this slide so pretty much based on You know essentially the sRIV requirements I mean, there's there's a couple of ways you can look at how you address Some of the needs that are coming from these V&F vendors Requiring sRIV so clearly the obvious answer is go build a cloud native V&F And personally for me, you know, I'm sure some some BU folks will be watching this YouTube video at some point You know, that's it's it's a message for them as well is if you can't build a cloud native app a really fundamental question of should you really be virtualizing that app should be asked of yourself and If if the answer that is yes, then you really have to put an effort to rewrite that application So it is cloud native And if the answer is it can't be then maybe the real answer or solutions that problem is automation You really probably have to automate it in a way that actually solves the real problem Because virtualizing something doesn't automatically make it automated Especially if the economics aren't there, of course a second a second option to that is Judicis use of sRIV and traditional cloud workloads. So what we mean by that is Alright, we'll drag you along kick in streaming. You know, you have your traditional via PNF function You've slapped it in a container or a VM and now you're saying you're you're virtualized or your your cloud enabled that's kind of a You know, that's a I was gonna use a word there, but that's kind of the wrong answer, right? I mean, I get it. There's probably some things that you just can't Avoid relative to finding the right people. It's it takes a lot of time a lot of investment But the barrier to entry you should remind these vendors that the barrier to entry is not removed So while we're reluctantly bringing you in and you're telling me I need sRIV for these level of high performance Other vendors now can come in and really differentiate with cloud native based solutions But you know up until now there are options to coexist these kind of workloads with cloud native workloads You know Moore's law, I don't know if you want to talk to that Toby or yeah So there's there's a number of options available to us And what I think is that in this particular area when you look at enterprise workloads There was a lull from like 1995 to 2005 where there wasn't a whole lot of Innovation happening in this in this space. It was VLANs and we're happy. It was 10 You know one gig interfaces. We're happy Now that we're pushing more on this both us and the hyperscale guys are pushing on this area The demand is driving innovation to happen and the Moore's law needs more things to work on so doing offloading of certain functions Makes sense. So instead of for me. It's an extreme SRV is at one extreme where we've done it. We decided we gave up We couldn't do anything in software and we just used hardware to solve the problem entirely That doesn't give us the extensibility or the software benefits If we go all software that we can't make it work quite yet So right now there is we think a middle ground where we have like smart nicks More sophisticated evolution of the existing nicks from Intel's and Broadcom's and such But then there's also the there's a new realm of smart nicks if you looked at what Microsoft's provided with Sonic or what? Others have tried net netronomes and others You're able to offload some of the standard packet processing functions and put them into an FPGA or a GPU And that we think is actually one of the ways that we can kind of Meet in the middle. I can take this one. I mean at the end of the day, you know, it's kind of hits at the hits at the point that You've again you the industry has built truly scaled out infrastructure like the internet I mean, it's it's the basis of pretty much anything we do in our daily lives now So, you know, that's you know, the buzzword is scaled out routing the reality to me It's in my mind given my background that's well, that's called VGP And of course naturally as a result a scaled out platform should look similar but not identical, right? So again, you don't want to reuse or sorry reinvent a lot of the tools that I've already been used to solve a lot of these Very large network problems and you kind of want to bridge the two worlds together and you know an example of a scale-out platform Is it's pretty much this right? You want to you want to get a sense of what problems you're solving in the cloud and leverage as much of the The tools that already exist out there to solve those key problems while at the same time introducing Through careful thinking and thought-out architectural discussions What what should be new in that new architecture to help enable the cloud environment as well, so One thing I'd add to this is it's more than just Taking something and making it a scale-out. It's also reconsidering what we're doing So if we take a packet through the out of the as I described before from the VM out of the building You don't want to really be opening it up and doing stuff to it that many times So there's if you rethink of it at a forest level, maybe there's a way to Consolidate routing functions or consolidate firewall functions with routing functions the or consolidate load Load balancing or or these types of thing that in functions together into one thing And a smaller footprint so we have to as is always the case as we move through this world in an agile way There's times when we iteratively keep adding and there's times when we have to step back and refactor and simplify I'll take this one So brownfield enablement again back to the physical vnps to make their lives easier during the transition so by nature Their legacy in a lot of ways and clearly they expect certain behaviors surrounding it So you can't just expect something people or functions have been doing rather for the last 15 and 20 years and saying Cloudify it right so there there's clearly some some vendors out there who will through business means should be incentivized to rewrite But at the same time you want to make the transition easier not only for for your benefit to understand the challenges as a result but also to to Evaluate certain types of functions and other options you have out in the industry and some of the technologies that are being used to enable this brownfield migration of Pnfs to vnfs in the telco telco cloud is you know again a lot of these a lot of these pnfs for liveliness detection Which is essentially a you know how do I detect the certain note or certain function is no longer functioning in a way that can provide the service You know mechanism like ping HTTP or I level checks BGP BFD These all you know I want to get into details like Toby warned you with acronyms But the net of that is these are mechanisms that all these vnfs are going to be wanting to use now So you have to make that you have to almost have a platform that enables that or else the transition just won't happen to even get the Ball rolling dynamic insertion network reachability. That's a very long way of saying routing and You know there's clearly some requirements coming from vnfs You know where they use routing as a mechanism to insert reach ability for certain types of tunnels that they were that they were Facilitating right and it literally is it's gotten to a point where you're saying well You don't have to do routing anymore just use this API to insert reach ability They're like well, I can't find the guy who wrote the app or my vendor went out of business So your platform needs to enable that for me. So that's kind of we're the guys that made that whole thing are all dead That's a possibility right so And then the two points I'll let Toby speak to because I know it's yeah So I mentioned the live migration piece and then another part that we're we're working hard on is Rationalizing okay, where are all these things gonna go? Where does it make the most sense and for us? It's been a you know one thing I would like to impress upon everybody for us It's an opportunity to take something that had been very siloed what direct TV or uverse was doing was very different than what voice was doing So this brownfield that we work in it has its it's fraught with all kind of minefields But it gives us an enormous opportunity to pull it together and so one of the brownfield aspects that we have to solve for is How do we deal with the resiliency when? The connection to your house is a single point of failure or the central office is a sent a central point of fail single point of failure We can't get away from people don't really have the option to have connections to two places So that that limits to some extent our ability to to do Cloud-native mechanisms as well the other part that I want to press upon everyone is the and then an opportunity for simplification Is that each of the p&f's the physical network functions of the old days came with their own orchestration tool? so there's Nokia orchestration tool. There's a Cisco orchestration tool. There's a Erickson orchestration tool that you know multiply that by a hundred So one of the brownfield things that we're working on is how do I take all of those things and transition them to a More common approach. There's one of the Things that we want to see the community do more of is how can I take a marano a mistral? Heat how can I take those open-stack components and bring them together to make and solve for? Quite a lot of redundant effort. So this last piece. I'll finish up with you know Left left to our own The natural path there would be I would have to attend four or five different standard boards standard groups efforts at us TM forum ITF whatever Oasis DMTF when you when the world's a telco and IT collide for the unit those worlds collide there are a lot of standard bodies Etsy So you have to attend all those and then and then now with open-source world. There's an ever-burgeoning set of of Open-source efforts that we go into so like not only open-stack, but open daylight and Like open contrail FDIO is coming. There's dpdk work opening that up. There's open data plane work. There's a lot of efforts So this is one of the reasons why I'm so I want to sell everyone on the concept of open FV there needs to be someone somewhere that brings it all together and helps to Coordinate the telco requirements and use cases and manifest that in something that can actually be used to test Are we gonna actually achieve this these requirements and be able to fill not only the functional ones But also the performance ones as I described before so in open FV We're doing everything we possibly can to document and clearly enumerate what it what we what we're needing and If and I urge everyone to look at it and then provide feedback Where possible to tell us where we need to be better Clearly when we talk about networking you can get lost in all of the acronyms and all the concepts and we often Lose sight of trying to make a much more simple description Sure, I mean, I guess the second one's obvious documented well entirely obvious but documented apis and data models You know that that's key right that literally gets everyone on the same I kind of think of it like everyone can speak English, but everyone's still either grammatically incorrect or talking gibberish You got to get everyone talking literally the same structured language and you know There's there's a lot of different efforts to do that. You know the world is kind of converge on certain data model representant representative type Languages, but at the same time that's all for not if everyone doesn't agree on what all that means So it's a Toby's point there. There's you know, one to the you know, two to the end exponential type operating Or orchestration tools out there that's mainly because they all speak the language with their specific applications or PNF slash BNF So I mean taking again a PNF and putting the VNF and still requiring your own own orchestration tool is not going to work I mean that just that's not going to fly right the reality is everyone needs to come together In some of these initiatives that document common use cases for the telcos and get everyone on board That's kind of my view Yeah, and then I'll close with just one last piece is that one of the really great things about open stack is where there was Not a storage standard for provisioning APIs. There is now one Whether it wasn't for networking there is now Enormous opportunity to make that happen And so we're spending a lot of time working with with the community to try to make this happen across many many different domains So I think that that really concludes our content for right now wanted to save enough time for some questions. So Yeah, all right, well just in general that You know as we talked about it's more than just the things that are Finding the right balance of hardware and software for simplifying getting the common APIs all that is very helpful to fulfill the cloud Requirements, but there's also as I was getting at at the beginning. We don't want to lose sight of okay The that's all commoditizing what we do. How do we add value? How do we make this into a platform that can evolve faster? So one of the things like what I'm working on lately is Adding storage to the story. How can I add storage to the central office to make it something that allows you to back up your mobile phone? You know, how do I add more? How do I combine services together? How can I take maybe IBM's public cloud or? Rack spaces public cloud and combine it together with our network and our our capabilities to make a hybrid cloud That's compelling and bundled These are things that I'm thinking about a lot lately is how do I how do we add more more value on top of this? Set of the base requirements. How do we make more requirements that make make us more valuable? What he said All right, so if anyone has any questions feel free to stand up step up to the mic And oh one last thing is we made the guy stop crying Thank you very much very interesting Do you think that there is an eventual desired outcome where the telcos will want to use a carrier neutral? Open-stack cloud purpose built for VNF Much like you see carrier neutral facilities being used where they are able to gain connectivity to the last mile Are do you think that AT&T and other telcos will want to continue to invest? Engineering resources to differentiate their own open-stack cloud That's a very good question. I think as it was kind of getting out with the the bits. I mean, I think Everyone here wants to see us find a way where all the telcos work together to make the bit moving the bits cost less and And I was getting at a little bit last night in our in our time together with all the different telcos Audit stack city, you know There is such an enormous opportunity to help get from the first billion people that have access to the next Five billion or six billion people and making this kind of technology available to a broader set of telcos Even the smaller ones, you know That's what the tip effort with Facebook's really working on right now And I think that's that's quite exciting and kind of it kind of I think gets to some part of what you're saying is that the carriers can work together to to sort of Make the bits less of important But again for us to continue to exist and be fruit, you know to grow we have to find ways to add more value Thank you Anything so be considered bias. So Yes, so a question we're talking about the technical challenges and moving into NFV and The question that I have is mostly on the moving to open source as a business model And the type of vendors that you expect will happen This is definitely a disruption because the current model of In which a lot of the equipment and gears that you're using right now is being sold in Appliances and very close kind of licenses model and moving to open source as a completely different model Do you expect that the new players that will come to that will be completely different than the incumbents Yeah, you're working with right now. Absolutely that this is probably one of the more important points I wanted to make as you've seen with with Unix just take Unix is a good AT&T example I mean for the longest time people made 20 like when I started in IT. I used to pay $20,000 for a C compiler That's a lot for something now. That's very undifferentiated and essentially is everywhere so it and then Red Hat and others have shown that it is possible to make a business out of taking something that Eventually gets out to everyone and makes a make can add value Themselves by adding services or support. I mean when you have to for us we have 50,000 red hat instances So they provide a lot of value just by helping us keep those up and running So there's there's always there's always something for them to do There's always something for people to add value on even if the bits themselves are Open-sourced now the part that I really want to compel everyone is right now many of the VNF's They aren't open source even with routing It isn't really a meaningful open source router that I can use it in the WAN and we'd like to see that We'd like to see that more with other other components and so I think you'll see that more But it hasn't happened quite yet. So there is there is business There is still a business to be had and Red Hat has proven it even if the bits changed open source So similar the question is to Juniper here. Yeah, I mean clearly we obviously are a company that has Traditionally offered a certain set of products over the years. I've been a proponent internally saying As the world ships to open source and even open hardware Who knows the stuff better than us especially in certain industries, right? So we can be the hot, you know the high-end services integrators For these types of services because look at look at Red Hat, right? They built a model around understanding and knowing Linux inside and out, right? Well, there's we have great guys, right? So it's a Boris's point, you know, we have a lot of good ninjas, right? So it's just time to lease them out for hire, right? It's it's a different business model And that's the biggest challenge for any any public company is trying to shift business models that way But I think it's it's a necessity. Yeah, and I also want to add make sure but he's clear on this open source Is not free that it was never meant to be free. There's still I am a developer I like developing and at some point somebody needs to pay me So there's even if I have apply an open source license to it so that everyone can get enormous Generative value out of it still there's somebody has to get paid for supporting it and the ninjas got to get paid Thanks Hi It was a very nice Presentation it's nice to see that there's a focus on the network because what the reality is network is our cloud, right? As as good as your network you provide better services But unfortunately from 1980s 1990s the network vision. Oh, I have private cloud I have I have a private network public network in the middle DMZ and The view is all these are to have has the same DNA same characteristics. Unfortunately. It's not true, right? So when you move into this today's world the front-end traffic versus back-end traffic there their DNAs are much different So what's your vision to differentiate the back-end traffic by means of providing facilities different views? For better services for machine-to-machine and to see better performance of the VNS. Thank you No, you go No, I go look so I'm trying to understand the question there. You're essentially asking how we see this this telco cloud Adopting to these different Requirements that you're seeing from different applications say on front-end back-end. Oh Someone kill this mic What I'm asking do you See a solution or features coming from open-stack for example neutron to allow you Do better features for the back-end traffic to handle more traffic because the back-end traffic is more Much ahead of the front-end traffic for machine-to-machine and help monitoring and stuff with the ecom architecture of the AT&T for example, there's a special Framework called OMF and DCA for analytics. This is all back-end traffic So back-end traffic versus front-end traffic the ports on the front end not necessarily to be used on the back-end vice versa So when you go at one security profile, it will not fly for you for a back-end for example Yeah, so I mean this is a very good point is like What we're trying to solve for is to solve this problem you're describing I mean and this is overlays are one way to solve it But I mean and neutron is certainly a key plays a key role in it, but this is why we want The network to switch to be more about software so that we can provide QoS in certain areas and not in other in others we can provide high throughput in some areas and not in others we can do full Full packet inspection in some places and not in others. We want to judiciously apply these techniques To the problem so we can be more efficient and software helps us to be that flexible to do that in where and when we need that Where the hardware wasn't able to do that in the past So now that understood the question I think at the end of the day we at least us as juniper We had we obviously had a very conscious effort to make sure we addressed all these issues as much as possible in the back-end But at the same time be very conscious to contribute a lot of that to the community as much as possible and that was the rationale for getting and Putting open-contrail out there because there are just there are a tremendous amount of requirements and this stuff's moving very fast I mean it it has to move fast because I believe it's an existential threat to a lot of the telcos So we're trying to keep the community open as much as possible By building the essentially what I call the implementation on the back-end for neutron to address a lot of those use cases And over time as those start getting pulled back into neutron in the community We obviously make sure to adhere to all those APIs and data models as they get defined and maybe at some point You know some of these projects do become part of neutron. That's you know, it's up for debate It's not it's not clearly in in my hands to make that kind of decision, but you know that's I don't I can't answer your question in a different kind of way But at the end of the day it all depends on the implementation to address those use cases And if you either wait on the community or execute on something you believe you need now With the with the expectation that it'll eventually evolve and become part of a larger community effort And that is our our ultimate goal as well. That's why we open-source it from day one and the beauty of open-stack And the ability to bring new things in quickly. Yep So you spoke about operators or vendors, sorry Not just porting their applications into Virtual machine and then call it cloud. You said you talked about how it should be native Could you explain maybe some of the things that you like to see done on maybe not the firewall or NAT level but more at the Higher level applications like the MMEs the HSS and the CSS these kinds of things what in your opinion would Mean that this application has been ported into into cloud and and it's native You want to like specific examples? Yeah, that's absolutely. So one part is very clear is that if I want to add more If I'm getting more load in that the way I add more Capacity is by scaling up and adding more virtual machines are writing more containers Whatever it is and scaling out that way and I can do that infinitely There's no stopping. There's no boundary and the pnfs. There's always a boundary So that's one one part of it. And there's also the thing where with cloud native you have an acceptance that resiliency of the six nines ilk or durability of the 16 nines ilk Doesn't come from being in one place It has to be able to be located in more than one two three times three places sometimes And many of the vnfs aren't made to do that and so Some part of it is inherent to making a connection and keeping a voice call up and running But we have to work through that and make it so that that can actually be serviced out of multiple locations to be able to maintain a high level of resiliency and a high level of Durability, so I mean that's one part of it That that we're asking for is having more flexibility scale out another part of it too with cloud native is is kind of working in an API model where you're Allowing yourself to be just a module within a larger framework and you're not thinking of yourself as the only thing And then reusing other components where there's where there's the possibility The crying guy at the beginning is crying in part because he used to own the whole stack And he could control the whole thing himself and that's also on our end on in our Organization we had teams that owned how I built the infrastructure and how I built the Presentation layer and everything between and they they don't have that anymore So going toward a more modular design is another another characteristic. That's that's unique to cloud native I mean that was probably my key point was modular design I mean I get there's a transition phase from going from a pnf to vnf but if you look at certain applications and In my mind if it's already running on some form of appliance, that's x86 base It just happens to be in branded hardware for a certain vendor That's ripe for moving it to a vnf But then you have to really really zoom into the application and the architecture this service is offering and whether you can Explode it out and really start rethinking how that fits into a larger framework That's that's kind of what I meant by What I said earlier All right, thank you everybody for showing up and thank you