 Hi everyone, it's work. Yeah, it's work. Okay So Thank you for coming. We're today. We're presenting on the the joint telco solutions red hat and Dell and My name Sanjay. I gary. I'm a senior solution architect for telco at red hat and I'm joined by Drew Schilke of Dell. Thank you. Yeah, Drew Schilke I work in Dell's networking business and I'm responsible for our NFV practice So today we're gonna talk about how Dell's solutions and red hat solutions fit together to build an open telco cloud and what are the benefits of Using our solutions. So with that Let's start with a little bit of introduction to Dell's telco solutions. Yeah So I'm putting up a slide here to kind of just talk about who we are at everyone came to a red hat session So just wanted to spend a little bit of time in terms of our approach and what we are working with red hat Philosophically very closely aligned in terms of how we're approaching virtualized Solutions for the telecommunications space What you're seeing up here is our purpose and belief statements that we've had for the past three years And I'm not gonna take you through it word by word In terms of what our corporate mission is but I think if you look Fundamentally is our design principles that we've had in place for over three years right now are very much aligned in terms of what we're gonna Be talking about here today, which is we always take a very compute-centric approach to everything We understand that software defined is the motion that this industry is headed in and we understand that everything needs to operate a cloud Scale and I bring that up because I think you're gonna see it come through here in a few in a bit in terms of the tenants That we're driving collectively in terms of our approach to things like NFP So if you advance on and we kind of think about our future ready solutions and what we're trying to bring to market with red hat you hear a lot of people talk about capex and op-ex in terms of what we're trying to balance out and You know in typical fashion you build a two by two quadrant and you want to be in the upper right and no mistake here That's what we're trying to do with this chart as well But I think kind of framing it up in terms of where we've come from a telecommunications industry and looking at the legacy in the lower Left-hand corner thinking about the monolithic systems in which hardware and software were tightly coupled together Drove a great degree of specialization especially around the vendor specific technologies in the lock-in that came along with it as We've moved into this space where things have become more virtualized and we can take advantage of things like cloud and NFV We've seen some of our competitors in this space move up into this chart and what we're calling new proprietary solutions Which is okay? Congratulations, we've virtualized everything but we're going to define every element of the stack all the way up through the top and You know to me it's kind of an n-minus one proprietary approach You're still going to get that lock-in penalty and you're not going to get as much choice Which is where we all have the opportunity here at the same time if I contrast it in the lower right-hand corner It's not like we can now just move to this model where you take a bunch of bent metal hardware throw it over the fence And things are going to work together. It takes more work than that it takes relationships between Hardware companies and software companies like Dell and Red Hat to put together validated solutions that we know Will deploy into the field easily that we know will work together and that we know we can drive the right interoperability ecosystem around it and that's what we're calling the future ready solutions in the upper right-hand corner We're trying to attack both the op-ex and the cap-ex problems Simultaneously and to this end I think we're philosophically very much aligned between Dell and Red Hat So if we take a look at what telcos Want to achieve in this environment? It's really around more than just Changing the hardware on which they're running it and running it on something else as they want to really bring in new technologies into play and that means for example Be able to turn up a new customer Very quickly not have a six-month installation time for getting everything working I mean that's it's the very basic thing that could increase velocity in terms of How telcos can operate and when we've talked to our customers we find that by and large They definitely want to do this. They want to move to a more flexible model And so bringing in the new technologies into that environment is Really something they're looking at now, and it's the right time in the industry to do that Yeah, and in addition to that There's this need for new services and business models that are coming into play and I think look no further Then what's going to be required in terms of new services as 5g begins to deploy and really really take hold As you open up the aperture of the bandwidth to those devices at the end that instant access You know mentality that most of us had today coupled with a greater level of bandwidth is going to drive a Need for a whole new suite of services that many of us have not even contemplated today And so what we try and think about in terms of all of our approaches here is thinking about the agility In terms of deploying new services and business models to go drive growth and revenue growth It's not just a cosplay. It's a top-line play as well And we like to keep that top of mind as well Yeah, and in terms of the competitive landscape, it's never been more competitive because for the longest time I in this industry, it's just been do what you've been doing but now with you know You service providers providing voice. I mean people are Making phone calls never using the telcos voice network, right? I mean, this is reality people are watching videos Without you, you know without using with just using over-the-top solutions And not using cable TV for example, right? I mean cutting the court very common So so now with that competitive landscape How do how does the telco? Become continue to be relevant in you know five ten years time How do they sort of bring bring that kind of agility back to into their their solution so that there can be more competitive environment so if we if we look at You know, where where where they're gonna start, right? They've started with some very basic scenarios of virtualizing what they already have We know that for example, you know, IMS have been has been virtualized More and more you're seeing EPCs With some some of the new players in the eval packet court Winning with software solutions You're seeing and of course the virtual routers being deployed for customers So that you don't need to put in a physical router and incur that additional cost and time so this is this is this is where this is where things start and as long as All of these all of these can be deployed in a standardized and scalable manner then There and they can be deployed sort of in a multi-vendor manner so that you're not tied to buying all of these components All of these software components from the same vendor That's where you can have That advantage of being able to plug and play and bring in new services very very quickly so I Think that's where That and that's where when we work with Dell We find that there's a way to really really test all of these kind of things all of these environments put them together And they're very good partner of ours in terms of validating all of this so with that I think we can Give you will give you a little bit of a sense of what is it that? Red Hat and Dell are selling From Red Hat's perspective What we sell is I mean everyone knows that we sell subscriptions for Linux, but Beyond that we have an open stack platform an open stack distribution and This is really now being Being tuned for NFB deployments, so all of the All of the Technologies are really being designed so that everything co-works together and this is important It's important to have solutions where you have different components that are tested and Validated together Part of that is of course our storage platform. We're contributing to the open daylight and Obviously our hype the KVM hypervisor is in there as well Now where it gets where it gets interesting is that? Beyond beyond that core, which we sort of illustrate down here There are a number of places where there's a desire in this industry to go beyond Just running virtual machines so With atomic host we have the capability of running containers and then to tie it all together There are a lot of environments where Middleware becomes really really important in a telco environment where they've got just hundreds of different processes services running and Tying them together becomes really important It's a challenge, especially if you bring in the Dynamic nature of having multiple Multiple vendors supply the core infrastructure then trying to make all that work at the top layer and manage it It becomes more difficult and needs a sort of a management system. That's very dynamic So with that Do you talk about del's components? Yeah, and the only thing I would add to it as well as we talked early on in terms of Driving the ecosystem and making sure that interoperability is very front and center in all of this And and this week you'll see an announcement coming out Of a joint effort between Dell and red hat and Tec Mahindra To stand up a laboratory in Bangalore in one of Tec Mahindra's facility That is going to be running exactly the platform that you see here With the goal of driving the ISVs that are creating all the virtual network functions that you see there in As an opportunity that they can certify on this platform and we can collectively start to build out that ecosystem around it as well So it's not just the platform in itself itself But it's also what are we doing to collectively drive people that can come in run on it? So that if somebody's trying to stand up one of the use cases we talked about earlier There's a known good starting point that we know things are going to work well Beyond that where we spent a lot of time in investing Is in making it a a confident experience in terms of deploying this platform? And and I don't know if any of you have had the experience of trying to stand up a proof of concept for network function virtualization but if you don't have a good set of ingredients and a good set of directions it could be a very very Frustrating task and so what we've done is invest in one the creation of of a reference architecture So there is a known good set of ingredients between Dell and red hat that we start with Very clearly calling out the versions that everything they need to be at and then we invest a lot of time in very detailed deployment guides from also the hardware from the open stack platform and then something we call the virtual infrastructure managers which is an element that we added from a Dell perspective To bring in some extra fabric management capabilities which are reflected in the lower right hand corner of the stack in terms of what's underneath it and running at all It's our standard x86 server base Which is capable of supporting both the compute and the storage aspects running stuff and then our Dell networking switches Which support our our our Dell operating system? But we can also pull in a number of our partners as well in our open networking approach where we've worked with a number of third-party Network operating system providers such as Cubilis and big switch and pluribus and so those are options that we can bring to the table as well from a network operating perspective In terms of how we can fulfill it and bring it to market We know there's not like one path by which these projects go And there are situations where we work directly with red hat to go to customers and have those conversations But there's also a lot of great many times where we work with some of the more traditional network corporate providers who have great relationships with these carriers and have a very strong services presence there and want to be part of these solutions that are deployed But then we also have a strong what we call an ecosystem perspective where we have partners in a management orchestration space partners that are systems integrators And again, this is what we think it we're all about choice between our two collective companies That's not just about a technology But it's also about in terms of how we can bring these things to market and deploy in a manner that you would feel comfortable with be it working Directly with us working through a network equipment provider or working through a systems integrator all options that we're willing to attain and all options that we have done and support of our customers So kind of tying all that back to kind of Some of the business priorities versus technical priorities. We talked about this need to deploy new services and scale at speed Just to kind of drive that home. I was at an event last week where one of the lead network Executives from one of the big carriers was speaking and they talked about the time at which it took to deploy a 10 gig or 100 gig circuit Just by maybe shouting it out What do you think the the baseline is to deploy a new 10 gig or 100 gig circuit in a very well-known tier one telecommunications company 90 days if they escalate and get everything to go, right? Not far off. It's four to six months is the typical range, right? How long does it take to deploy a new piece of network equipment in the central office? Two years. Yeah Yeah So so this is what we when we talk about in terms of you know the agility and what we're trying to go in terms of deploying new Services, it's also the infrastructure behind it and being agile on that But coming along with that we also know that there's a pivot of employee skills that needs to take place as well and people have a lot of time and and an equity and certifications on certain equipment and certain equipment from certain vendors and This is a very very different approach which is going to require a skill set pivot and and there's no great easy answer there As well, some people are going to have a hard time resisting that You know, even if even if firms take a lot of time and try and drive that training that's necessary to go do it Kind of our experience in terms of what we've seen good success in the field is when the carriers take the resources that are in this and Involve them from day one in terms of defining what this new virtualized network function is going to look like And have them involved in design and the proof of concept and deployment as opposed to waiting to the end to bring them over But that's just the business priorities. There's also the technical priorities. Yes so What we've found that really to Develop an open telco cloud and modern open telco cloud They've got to modernize the way that the applications are are developed and deployed. I mean, I think we all know this Moving more to a cloud native type of application development and this is this is a big change It's not just taking what's already there and virtualizing it and putting it in the cloud so That's that's really, you know, that's really the first the first the first piece of this It's absolutely essential that everything be kept available that goes without saying in telco environment but availability means something different now because Used to be that five nines meant that your hardware had Extra redundancy and you had all kinds of fancy technologies in the hardware to ensure that now It's about service availability and how do you keep the service up regardless of What's happening underneath and this is the way pretty much every cloud scale environment is running Can how do we how do we bring telcos along to also develop along those lines and finally? For this to all work, it's really got to be open open for plug-in of different types of capabilities and technologies and It's also got to enable the business priorities Which means we need to be able to do more than just what's already been done in telco networks That means be able to gather data from the network do some analysis on that data feedback and Dynamically change the way your network operates rather than wait for trouble tickets to come in So with that I want to leave you with a sort of a vision for an open telco cloud that we we've been talking about at Red Hat and what this involves is is thinking about the application as More than just a virtual machine. It's really Taking the open stack platform realizing it's not just open stack But you've got a network layer underneath it that the telco has deployed You've got in fact most of these operators have mobile networks and more and more applications are being Mobile first so that has to be a key part of it How do you enable security all the way down from here all the way down from the mobile access layer up through? the Open stack platform and even up to the application the application is not actually running on open stack the application is a collection of Scaled out virtual machines or containers that are tied together by a Middleware platform, I mean that this is how it's worked in enterprise for years and years We're now realizing You know telcos are realizing that they've got to start developing applications this way as opposed to the one application per box methodology Or one application per VM methodology even it's got to change to this type of environment and so What we can offer is Consistent support model for the tools needed for all these components Whether you're using the various components of JBoss whether you're using the various components of open stack Whether you're using Seth for storage We're using rel as an operating system, and you're using Dell hardware and the Dell fabric all of these can be consistently supported And you can plug in whatever functions you'd like to plug in I'd also like to Call out the fact that new business models and new operational models can be invented once you do something this way because Now if you have extra load on your network and your network can feed back events back Middleware events essentially back up into your application your application can adjust its behavior if you think about Video streaming for example video streaming is something that a lot of times It's done today by flooding the network with as much as possible and then seeing if it worked And if it didn't work and you say oh, well, I get a down res that video, right? Well, what does that do the network if everyone's doing that is you're flooding right? So it'd be better to be able to have a feedback mechanism what? So what we're talking about here is creating an infrastructure whereby this is possible. It's not Solving and saying that oh we already have software that does all this I mean it would take a lot of companies to build all this But what we're saying is this is the basic infrastructure that would enable That would enable developers to build these types of new applications that leverage That that leverage these capabilities and enable new business models and new operational models so with that We're happy to Take any questions If you don't mind you can step to the mic because they're recording it We want to hear what you say don't worry won't film you just want to hear you Hi, this is Fatih from Ericsson Yeah, the main point at least for me going towards the cloud is the Utility-based competing right meaning You will pay as much as you use when the things comes into the service provider or the product provider side to the Dell and Red Hat especially Red Hat There is nothing new in red hat side for by means of licensing if you look at even if you try to use red hat OpenStock distro. It gives you 30 days of license for two sockets Even there is no open source version like in the centOS case when you go by means of scaling up You have to do the ups, you know upstream or Preparation of investment before you utilize this infrastructure. That's one question Is there any any new licensing methodology that you foresee in an FBI? For Red Hat there are my second question for Dell are you guys really collaborating by means of beyond of certification? How about optimizing this stuff right when you do the CPU pinning by means of virtual machines? How about how about the worker track pinning in the CPUs? Is there any any effort that you plan to do? Thank you so So in terms of in terms of licensing. I mean look this is Red Hat's been Creating you know a lot obviously talking about NFV You know and bringing in a lot of new focus into this effort over the last year and certainly in terms of in terms of different types of subscriptions and we you know We don't really call it licensing we you know We refer to it as a subscription and subscription model if You know it is a change to move from a pay per CPU Essentially or per socket pair model to a pay per usage model Having said that there are a number of ways that we work with customers on these on these subscription models And not every you know It's not always locked in to the model that was stated in the question now You know, but those are like every company there are there are individual deals that can be worked out On the second question you're talking about CPU pinning specific. I didn't really get that CP pinning specific to Dell Okay, so Thread and process again again pinning pinning to a individual CPU individual CPU core and So I'm trying to understand where where that particular question what you're asking of Dell in that in that question If you cannot do the pinning Yeah So so first of all first of all, let me explain the current state of this is with red hat OSP 8 We're still operating with virtual machines right not containers So let's be clear about that so we're not claiming that and there's a reason for that because we if we don't have something That's delivering what you're talking about then we can't come out and say hey We support containers in the V right we're talking about so the containers in this presentation It's presented clearly as a vision that we want to get to I mean obviously everyone's get there But as red hat we have to make sure that what we deliver Does provide what you need to actually run the stuff and if we can't do that yet Then we're staying where we can deliver which is with virtual machines Yeah, the only thing I said is in terms of just the work we do collectively together It's not just the validation we actually have a dozen developers at Dell that worked on red hat Basically as an extended team of red hat In terms of the kind of things we work on though I'll be honest the majority of the focus lately has been on deployment capabilities That's an area. We think we need to go off and address in terms of what we would do to make the Dell hardware better We would try and be very careful to balance that we wouldn't want to do anything Bespoke to us because I then I put this in the lower in upper left-hand corner of the box that we said we don't want to be in right And we want to make sure we're always working with red hat to drive it back into the primary distribution So we're all on the same page on all of that But that's kind of where our focus has been lately on deployment reporting and manageability This is a safe from Nokia So the same this stock is more from an FV point of view So my question is more specific and more from a Dell point of view So and first thing is this is this is great initiative. So that definitely this is Area, so my question is that are you guys are considering to use any other platform hardware platform other than x86? Which could be more friendly with the network functions Are you considering on the compute side or the on the bare metal on the on the lower level because you are providing the hardware platform to Understanding yeah, are you considering anything other than x86? I mean we we Consistently look at alternative architectures and And I was actually part completely coincidental here. I was a part four years ago We announced a proof of it was a proof of concept system We didn't create revenue off of it with a 32-bit arm processor and our server We called copper and we made it available to labs and ISVs to go off and play and test around with it And so we consistently look at that and we still have more POCs that are non x86 that are out there on today But to date we've not announced anything in terms of a product level or you know Anything we would go off and sell and and create revenue off of other than x86 So that's that's a statement of where we are today Will we always continue to look at those alternative technologies? Absolutely But x86 is the platform that we're on today from a cute perspective So very same question for a from a redhead point of view Are they considering anything on the top of the red hat Linux that that could be helping to be more? More performance wise better when you somebody is running the network function like a virtual router You gave an example use case. Yeah, so running that one on a virtual machine Which is not maybe friendly to the network function are forwarding and those so are you considering to add any layer or changing in the The way right now the status is that it could be more efficient in future. Yeah, so actually that's a good question because Very timely because just just last week. I was on a panel at the at the Which conference was that? In a few and every world Congress. Sorry. I've been to so many things but I was I was on the FDIO panel and and So there is there is an initiative where You know many players in the industry are looking at in in use cases specific use cases where they need that type of Environment that they could leverage things like Intel DP DK to get better performance and that is you know that is supported But also understand that that's not something that's not a silver bullet that you put Everywhere and say well use this use these technologies everywhere because then you know by and large most applications Work well by running on the standard with the standard Linux kernel networking But there are certain applications that you could get better performance by doing specific things Yeah, I mean your question was kind of hardware and performance just There's other things to go look at outside of your underlying CPU architecture There's I mean there's some companies doing some really interesting things around SRIOV offloading open virtual switch offloading right which basically are delivered in the form of a PCI You know express compliant Nick right so you're not having a kind of fundamentally upend the architecture Effort to go get what can be you know orders of magnitude gains and performance for a given use case Off of all that so I mean that's if you've not looked in that that's an interesting place to go look as well I think you're going to see a lot of interesting things coming through there In conjunction with FPGAs that could really optimize on the hardware side any other questions Okay, okay. Well with that. We thank you for your time