Steve McDowell 0:03
Welcome to another data centric Podcast. I'm Steve McDowell and I'm here with Matt Kimball. Hey, Matt. Hey, Steve,
Matt Kimball 0:08
the Silicon Valley edition.
Steve McDowell 0:12
So we are road tripping in Silicon Valley. But our guests are not our guests are in Austin. So tell me, Matt, what are we talking about this week?
Matt Kimball 0:20
So what we're talking about. So, you know, we talk to a lot of customers about their journey to the cloud, the challenges they have with the edge and modernization in general. You know, how you how you deploy, provision and manage infrastructure and manage your environments, kind of holistically, and how much of a pain in the neck that can be right. Deal with it every day. And so, you know, we've we've talked to quite a few customers, and, you know, kind of some of the conversations that I've had, with a lot of the, the solutions folks in this in this space have been interesting, but one of the more has been with, with Cisco. You know, it's funny, because you and I, and we think about Cisco, and a lot of folks in it, think about Cisco and immediately go to their heads go to networking, or it goes to UCS, their server solutions that are that are quite powerful and quite popular. But there's so much more to what Cisco is doing right around management within our site and around HCI. So we wanted to bring in someone who knows everything about anything. A rush fan, a Cisco senior director, Todd Brandon, he's thots. The is the Senior Director of cloud platforms. He does a lot of product marketing, positioning, but he's been in the space for a while. He kind of gets all the dynamics with with the customer space. So that'd be great to have him coming in. Have a conversation with
Steve McDowell 1:53
us. You're talking about him like he's not in the room. Hey, Todd.
Todd Brannon 1:57
Gentlemen, thank you. That's some that's some high billing about everything. But then on my LinkedIn profile,
Matt Kimball 2:06
I'll probably do as a reference for you. How's it going?
Unknown Speaker 2:09
Never better. It's a beautiful day here in Austin. We need to do this from ski shores next time.
Matt Kimball 2:14
Oh, no doubt, no doubt. We'll actually be down there in the next couple of weeks back and back down there next couple of weeks. But yeah, Hey, Todd. So it's funny because Steve and I were talking last night for all the conversations we've had with you. And with Cisco in general. I think this might be the first time we've had Cisco on the podcast, and we're terribly sorry for that.
Unknown Speaker 2:37
So happy to help correct it.
Matt Kimball 2:40
Well, let's make this a running conversation. Hey, so But before before we get into like the meat of this, give it tell us a little bit about yourself, Todd, kind of where you know, what you've been doing, where you come from, you know, Cisco has been like,
Unknown Speaker 2:54
you know, I saw I started life in the army actually a long time ago. And I was an armor officer, so a tanker. So I had a lot of fun. Just leading troops was an incredible experience. And then when I finished up there I was looking for for a job. And I was here in Austin, and Dell hired, Dell hired the smartest 3000 People in Austin and hired about 30,000 more. I was I was in that best sucking sound going into Dell back in around 2000, or crushing 9495. I started there. And that's how I got introduced to servers and storage and workstations and kind of grew up a first job there was running benchmarks and found my way into product marketing. And then just kind of, you know, Battlefield promotions as the business grew type of thing. And in 2009, I was working in a group there that designed custom platforms for the cloud providers. So before they really went ODM director building their own, they were still buying servers from the likes of Dell and HP and others. And we had a really fun little business that did custom designs for, you know, the folks at Azure and Facebook and all the rest and but while I was working there, Cisco decided to get into the server business. And I was tasked to do competitive analysis on what they were talking about doing. I said, Well, that looks pretty cool. And so a few of us from Dell came to Cisco and then you know, folks resign and HP and there was a tranche of us that all joined Cisco back in oh nine to help the company get into the computing market. And I've been here ever since it's just been a real blast. I love the company, fantastic leadership and I'm really proud of what we do around it.
Matt Kimball 4:39
You know, it's so neat. It's funny that you went from kind of competitive analysis to fan to employee that's a green
Unknown Speaker 4:50
light with so much I bought the badge
Matt Kimball 4:55
Hey, so you know, for our audience, you know, obviously, you know, we speak to it, folks. You know, administrators, directors kind of up and down the food chain. And as I started in the intro, when you were in the green room, you probably heard us talking. You know, there's a, you know, there are a lot of challenges that it is facing today. Right? And it's not, you know, there's the promise of the cloud and the reality of the cloud, there is the promise of digital, digital digitization of your business. And then there's the pains of going of trying to get on the other side of that transformation, right? Is this what you hear it at Cisco, as well, when you speak with your customers and get feedback?
Unknown Speaker 5:35
Absolutely. I mean, the spectrum of adoption is as varied as you know, all the different customers and industries that they're in, right, and you've got, you've got kind of two minorities, I think, at either end of that spectrum, you know, you've got the folks that are like, Hey, I'm running on prem, and I'm happy as a clam. Don't talk to me about cloud cloud, they're increasingly a smaller, smaller minority. And then you've got folks at the other end that were either born in the cloud, or quickly got there and are 100% cloud native. But the vast majority are of the customers we work with are in that messy middle. Where they're working with usually, what we see is they're working with one primary cloud provider to try and get to some centralized kind of operational consistency. But inevitably, they end up working with multiple providers. And it's, it's nobody has a strategy to go multicloud out of the gate, maybe someday, right? Where they say, Hey, I want to have multiple suppliers to try and manage costs. But that really doesn't pan out, because most of the cloud providers give you increasing discounts based on how much of your footprint you allocate to them. Right. So the cost piece doesn't often work. But there's different reasons why people might have strategically, you know, adopted or, you know, entered into operations with multiple cloud providers. But most of the time, it's an accidental journey, right? Where there's like, Hey, we got to move into a new geographic region, or we have a, you know, very specialized requirement that takes us to a particular service on a particular cloud, you know, so most customers are in that situation where they found themselves in a multi cloud hybrid cloud operating model out of necessity, but they never really built the strategy for how they're going to operationalize that. And so yeah, that's, that's what we're seeing just kind of a really broad range of most people in that really difficult position and trying to figure out operationalize
Matt Kimball 7:30
it. So how do you as a, you know, as a, as a, you know, I'll refer to Cisco as an IT solutions provider, because you do more than, obviously, I mean, you run the full gamut from networking to, you know, to compute all the way to all of the software you do on the management side? How do you? How do you approach a market where there is no, I mean, every every customer is a unicorn in a way, right? Or a snowflake in a way? And that in their journey to the cloud and their adoption? And how do you how do you kind of come up and say the, this is how we serve the needs of the market in general? Well, I
Unknown Speaker 8:07
mean, I'll be the first to tell you, like, you know, when when people think of cloud technology, companies, we're not one, ironically, we're not one of the first ones that comes to mind often, right, the AWS so maybe, you know, the big tooling, like se hace corporate TerraForm are some of the first things that come to mind. But ultimately, you can't have cloud without networking. And when it comes to networking, we know a thing or two, because we've seen a thing or two, here at Cisco. And, and so we really approached each customer, kind of the way you described right, understand, where are they? And what are they trying to accomplish? Are they are they trying to, you know, expand or develop, you know, into new new markets? Are they trying to expand into new geographies? Or they're, you know, they trying to adopt? Is their strategy driven more around an operational set of questions like, hey, we feel like, we shouldn't be in the business of managing our own data centers, we want to get this to the cloud. So customers come at it from a bunch of different angles. And like you said, we have a technology footprint that is so broad that from a storytelling perspective, which is a lot of what I enjoy, you know, trying to figure out here at Cisco, it almost collapses under its own weight, because we have technology that ranges everywhere, from connecting users and devices, to data and applications on the other side and on the cloud, we and everything in between. And so it's really that that's where our sell our sellers and our our, our solution partners in the channel, really earn their their money, right in that in that consultative approach and figure out is what we don't want to do is walk up to customers with a kind of one size fits all strategy. And when and when I see you know, the the industry's kind of presented customers with two basic choices right? In terms of how they approach it, it's like, Hey, we're either going to help you lift and shift your on prem operating model to the cloud. And that might be most exemplified by say, VMware his approach to the problem. Or you've got it from the cloud provider side where they say, Hey, we're going to help you bring the public cloud into your dataset with a solution, like outposts, for example. And those have a lot of merit. But they they're also a little bit inelegant, in a way. Yeah. So everything that our engineering teams are doing here at Cisco is really kind of focused on how can we, how can we create operational integrations that work at a more fundamental altitude, right. And maybe an example might illustrate this, right when, when we think about how to do cloud networking, for example, one of the technologies we have is our ACI framework around data center and cloud switching. And rather than take, you know, an existing environment of ACI try and just plop it in the cloud, what we did is we did API level integrations with all the major cloud platforms so that you take your policy that you've developed, and you can translate it into the the, you know, the very atomic knobs and dials that you would turn, so that you're not imposing a new networking schema, and Amazon or an Azure or Google, rather, we're doing kind of a universal translator, where we go out and use the API to dial in the networking paradigm, you know, within that cloud provider to get to the end state you want. Yeah, without you having to necessarily learn how they do. Because, you know, all the providers have a slightly different approach to networking. And, and rather than have to learn all those, and rinse and repeat, we act like a universal translator, almost for policy. And so it's at a more fundamental primitives kind of level, right? That we try and that we try and do our engineering, and with a very cloud agnostic approach, because we got customers are all over the gamut of who they want to work with. And so we don't really want to take, we don't we don't want to, you know, have a favorite, if you will, in terms of which cloud providers we're, we're engineering to it's, it's across try be very agnostic there. So I think that's what's a little bit different about us is we're trying to work at a much more down in the primitives, and kind of help customers operationalize in a more, yeah, maybe cloud natives or a highly charged word, but in a more cloud native way, right?
Matt Kimball 12:23
Yeah. Well, I think I think you bring up a good point there too, in it from a little bit earlier on, and that, you know, when when you talk about, you know, their their, their, their extremes on on either end of kind of the cloud adoption continuum, if you will, most people live in that messy middle. And amongst most companies, and, you know, the in the ability, the ability, or the the capacity for IT organizations to fully optimize, you know, their on prem to a multi cloud environment is next to impossible, right, you don't have enough people, you don't have enough of the right skill set, you don't have the right technologies, and to what you said a lot of a lot of IT solution providers, you know, say here is our solution, make it fit in your, you know, and, and they try and build something that suits the needs of everybody, maybe not quite as well, whereas it feels like Cisco is certainly more more understanding of all of the phase of issues facing in, you know, like Tower of Babel, right, that allows me to talk to any cloud in a very in a much easier fashion. But when you get to the operational side, it's one of the areas I want to jump into, because we've talked to a lot of companies about this, the operational side and managing. You know, one of the things I was impressed with, and I'm always impressed with when I walk away from our conversations, is what Cisco has done with inner sight, which I don't think gets an nearly enough attention in the market. Can you kind of, you know, for our our listeners, who may be, you know, aren't familiar within our site, can you give a 10,000 foot level?
Unknown Speaker 14:03
I appreciate you bringing that in? We, you know, we the reason I move to Cisco was how I saw the the management paradigm being developed around compute. And, and at that time, you know, things were very Device Centric at that moment in time. And so you had management modules and blade chassis 's and you had you had all these different management elements that would be spread across the data center in what Cisco was trying to do is consolidate those in the network fabric. And in that threw off a lot of benefits Policy model for all that. And so that was where we got started with Unified Computing System. And so we drove down that road in about 2015 2016, we realized we need to take this control plane from the network fabric to the cloud. And so that's how intersite really came to be we you know, we launched it in 17. And initially The, you know, the the value that we were providing with intersite was related to infrastructure on prem. But the value prop of moving it to the cloud was just all the benefits of a SAS approach. So, hey, we can connect your operators anywhere with your infrastructure anywhere. So Riot Games, right? They do Liga, legends and esport. Gaming, right? They've got 12, I want to say 12 broadcast studios around the world, where they bring in gamers and do tournaments and broadcast it to the world. And in a scenario like that, you know, you don't want to have to have hands or talent in each of those eSports facilities that know how to configure the servers that the games are running on. So right, so So riot came to us and, and we suited him up with nexus for ultra low latency switching in these eSports arenas, because that's obviously, you know, really important to gamers and their tournaments. And we have UCS compute. But then within our site, they were able to look at their all 12 of their Esports arenas around the world. And that all that compute in those in those arenas. And from an administrative perspective, it looks like it's in one datacenter. And so if they want to do something like hey, and this is really tactical, but it kind of shows the day to day, right, if they want to disable cores on a server, so they can overclock the other cores, and soup up the gaming experience. Rather than go in and have local hands doing that machine by machine around the world, they can set that very granular policy element into the definition of what it means to be illegal legends server, whatever their game files, their their gaming, and then they can they can apply that uniformly to all their machines around the world. They're doing it from the cloud. So our value prop with intersite was originally Hey, what is SAS managing infrastructure as SAS from the cloud? How does that benefit you? The other thing it did was, it got us out of a mode of chasing our customers around saying, Hey, have you upgraded to the latest version of UCS Manager, right, etc, etc. Now, they're just logging into intersite. And any new features we deliver just show up. And they don't have to think about building and maintaining a monster to manage their monster. So that's where we got started with with UCS. And we did some things like connected tax. So we're kind of in the data center with our customers where they want us, and we can alert them to problems and we automate, you know, problems that we identify, we give an automated remediation. So a lot of things that we're doing are just trying to take, you know, day to day administrative burden off of customers. And so that was the maniacal focus initially. And then we realized, okay, you know, infrastructure today no longer means hardware on prem, it means the abstracted layers, right? Virtual machines and containers. And so we started to build, particularly kind of early last year, we started building out services on top of this platform that kind of address that. And so we started with some things around containers, we have container service, we did some workload optimization, so you can decide where to place workloads on prem or in the cloud, and lower your closet, there's like three or 4 million SKUs instance types, right? Across all the cloud providers. How do you keep up with that? So we have an optimization service that is keeping up with that and says, Hey, you're on instance, type A, based on your workload profile, we think you should be on instance, type B, would you like us to do that and save you a buck or two? So we've kind of been moving up the stack now, out of the on prem infra, into these elements that span hybrid cloud.
Matt Kimball 18:39
I mean, that's, that's killer, right? Because it for a number of reasons, right? First, first, is it to me, you know, when you think about the early days of intersite? To me, that's the realization of what, you know, large enterprises used to try to achieve through their Knox, right? I mean, and there was always that notion of single pane of glass, but it never really kind of was realized in a real and simple way. But as you move up that stack, you know, it sounds like you're almost talking about managing all of your cloud operations and, and acting as somewhat of a broker, or at least a suggested broker to help, you know, not just operationally but economically as well, for these IT organizations.
Unknown Speaker 19:21
Yeah, that to an extent, definitely. You know, and I would, you know, I, I really shy away from trying to portray intersite, or any platform really as the one platform to rule them all because I was with a, I can't name the company, but I was lucky enough to be in a conversation with a CIO and some of our executives and, and she said, Look, this is how we basically operate it today. That is, the business comes to us looking to create a new experience, right, whatever that experience may be. We go off and develop a solution that answers that need to deliver that experience. But then the next step for us is to look outward. In the industry of platforms that we can then integrate to create that solution, right? They're not building it from scratch. So that that's what's driving that, that that paradigm, right or that that approach that the IT teams are using today, is really driving our engineering and innovation agenda. And in the sense that we're taking all the technology at Cisco, that used to be in a lot of disparate systems, or embedded in products. You know, step one is let's salsify, those things and move them to the to a cloud platform. And then let's aggregate them into logical buckets, based on, you know, the users that are going to use them because again, you may not want to aggregate all the stuff you've got, you just want to kind of get it into the right chunks for the operators that care about security or net ops or, you know, computing infrastructure. And so our platform strategy at Cisco is really aimed at how can we aggregate this technology into easy to use platforms, and then create integrations between them that help our customers, you know, bring us all together. So integrations between platforms that share telemetry data, for example. So you've got two operators may be using App dynamics, for example, to look at the application topology, you know, we acquired AP, the A few years back, fantastic visualization of the application environment, well, we integrated that data into intersite. So that not only can the users using App D, see how applications are connected down into infrastructure. But now the infrastructure cats using intersite can see how the infrastructure that they're accountable for, for operating is tied to the application and all the distributed components of that application. So those kinds of integrations, help our customers get to, you know, a single source of truth between teams. So, you know, instead of feeding revenue to the local pizza shop, by having all night, warm sessions to root cause problems, you know, it's on us, we think, at Cisco, to bring a lot of this telemetry data together behind the scenes, to give a rich experience to different platforms, but have it all feeding off a single source of truth, a common language. So it's very easy that that's the biggest challenge people have is just seeing how everything's connected. And that's what we're trying to discuss how can we help customers connect the dots and give them situational awareness? That's what we refer, you know, full stack observability, right? It's a huge topic in the industry, kind of moving beyond monitoring. But really understanding how the pieces are connected in real time. So you can get ahead of problems. And that's really where our focus is, is this platform strategy and the integrations between them. So we can help those CIOs stitch things together.
Matt Kimball 22:43
What are your views? And I know this is a little bit off topic or not really, but what are your views on kind of autonomous, you know, heading down the autonomous infrastructure route me because I think about what you're saying, right? And, you know, you have you do have the tool chain has support full stack observability. And he talked about AppDynamics, intersite, and all of the other pieces in your portfolio. And it seems like you know, that the next natural extension is not just hey, you can see what's going on. But these problems are root caused and resolved before they even happen, right. Predictive failures or, you know, whole story. Yeah. What are your thoughts on that?
Unknown Speaker 23:25
I'm smart. I know, we're an audio podcast here. I'm smiling. Because, you know, I T operators are paranoid, unnecessarily paranoid and conservative. Bunch, right? I mean, they're the ones who get lit on fire when stuff goes wrong. So the idea of, you know, you know, autonomic operations is been around for a while, and there's definitely, you know, ways that you can, you can do that. But every customer I talked to generally says, look, keep me out of trouble by alerting me to things, right. And then give me the easy button to fix it. Yeah, but I want to be in the middle. Right,
Matt Kimball 24:05
I want to press that button,
Unknown Speaker 24:06
I want to I want to be alerted. I want to press the button right now. Now. Now. Now, what's what's starting to evolve though, what I what we see starting to happen, though, is maybe low stakes decisions. They're willing to fully automate. So an example might be that instance type scenario, where, hey, you know, instance, a is going to save you a couple bucks over instance, B, they're both on AWS. You know, you're not going to lose, you know, nothing's really you know, there's nothing structurally changing in your in your overall topology. So maybe those are, you know, things that I'd be willing to flip, flip the full auto, if you will, but I think it's graduated in that way. So low, low risk, low risk elements. Yeah, some people are starting to adopt the full automation strategy before. I think the mainstream IT teams, they're still in the mode of I want to be in that decision chain, but many Make it easy for me on both sides, right? You know, you know, help me separate the signal from the noise with with good analytics, help me connect the dots, and then give me the easy button to fix it. And that's what I see there. It's very, very, I haven't met a customer that is like, hey, put me on full auto forever that
Matt Kimball 25:18
oh, yeah, no, it's funny, because it's, it's a story we all like to tell. Right? And it's, I mean, as an analyst who, you know, we can say whatever we want not be held accountable. Even though we haven't
Unknown Speaker 25:30
marketing, it's the same thing.
Matt Kimball 25:34
Very, very thin line between you and I, my friend. But you know, it's a, it but you know, it's easy to kind of build these grand visions. But you're right. I mean, concern it and it has a right to be conservative and their adoption of technology. And it is certainly a slow moving train. You know, what's funny is we are we are 30 minutes into this conversation. And we haven't we just got started. I know. And we're on with Cisco, and we haven't talked about hardware at all.
Unknown Speaker 26:05
Oh, man. I know, there's some some server augers out there that one of your slots and watts will have to come back?
Matt Kimball 26:11
Well, but But before we can, before we drop, I want to kind of give us a little bit into you know, it. It's funny, I say that about, you know, we were on for 30 minutes. And it's been a software discussion. But, you know, Cisco is a is a is an evolving company, in my mind from kind of where they were even just a few years ago. And it seems like, you know, you're one of the few companies that seems to understand this evolution of it, and the the weird journey that it's taking to the cloud, right, that messy metal and trying to meet folks, but, you know, you do have a good hardware portfolio. And there are a lot of different needs out there with with edge with your traditional compute with, you know, AI and the higher end workloads. You know, I know you've done some stuff around HyperFlex and UCs X, do you want to
Unknown Speaker 27:03
know? Yeah, absolutely. Right, I think that's what kind of sets us apart in some ways, is we're innovating up and down the stack, from silicon, all the way up through the things we've been talking about in the operational model. And where we still see opportunities to deliver value through silicon innovation, we're still doing it, we just released our latest incarnation of our virtual interface card, right? That's a custom ASICs that we build. And it's a converged adapter that you can, you know, you know, it can transform itself to rep, you know, appear to the operating system and appear to the machine that it's hosted in, as you know, 256 virtual adapters of any kind, you want to build Ethernet, or HBA. And to do that, you've got to develop some custom silicon to deliver that kind of capability.
Unknown Speaker 27:53
The other thing that it kind of dialed in on computing, the computing architectures have really changed dramatically, just in the last five years, right, it was kind of quiet there for maybe a decade where you had your classic food groups have, you know, processor, memory storage, but now storage is coming a lot closer to compute with NVMe. The, you know, AMD and Intel have done some incredible things with core density and options there. So, you know, the architecture, how you build in fabrics, right? Cx L, you know, you know, PCI, and where that how that's evolving the fabrics and how the subsystems of a server are interconnected. There's a lot of different ways to, to approach it today. And so we, you know, we've done some things there with this platform introduced last summer called X series, and we kind of looked at the problem, I won't say, you know, composability, right has been this, this, you know, this this vision or Mirage, I think, on the horizon for a long time in the computing, architecture circles, but there's some things starting to happen where we can start to disaggregate components and then assemble them, you know, the way we want, you know, from, you know, through a software control point. And we're doing that with our X series product. You know, in our announcement this month, we, we introduced basically a PCIe blade, right? So you've got server server nodes in that in that X series chassis. Now we have a node that plugs in that you just populate with PCIe devices, like GPUs, for example. And so we've introduced this architecture, we have a, you got a classic converged fabric that I kind of talked about, you know, for Ethernet, San management traffic, but we added a second fabric. It's really we call it x rabbit, but it's it's for the purpose of connecting server subsystems together. So now we can take a workload that you would typically have gone out and maybe had a specialized system for maybe you had an external array like that, you know, there are vendors that have the GPS You know, boxes basically, and then you know, cabling hooked up to your servers, we've kind of streamline that with our x fabric so that you can assign resources from that GPU blade out to that server nodes and, and do that through policy and kind of assign the GPU, which is a very expensive resource relative to others in the system, you know, assign that to the nodes that need it when they need it. And so, so yeah, there's a lot going on in server architecture. And we're having some fun with it. Because we're all about fabrics. We're all behind connect stuff. And we got buildings, real engineers, are good at that. And so our server architecture is still over the horizon. I mean, we think we made some big waves when we introduce UCS, back in Ohio, and now we built built our, our chassis Foundation, our architectural foundation deck series for the next, whatever, 10 or 12 years and, and by, and one thing we're trying to do is decouple all these subsystems, because they all they all kind of innovate on different cycles, right? You get you get innovation from different components in the industry on different cycles. So we want to make it all modular so that, hey, in two, three years, like today, our fabric is PCIe based, maybe we want to use a different fabric and two or three years, it's just, you know, swap out those x fabric modules. And now you're on a different fabric for the interconnect for all the subsystems within within the chassis. So yeah, we're having a lot of fun with with baseline architecture at the product altitude. But then the magic is how do you connect that up into the operating environment that we do with, in our case, intersite. So that you're not creating more complexity? That's the key, right? Like the big problem. composability has been, you know, who wants to have to deal with the complexity of how to put all this stuff together? How do we make it automatic transmission versus a manual? Right? And so, how you tie those physical architectural decisions and opportunities? How do you make that easy for customers to consume the benefits of those things? And that's where we focus with a control point within our site and doing that from the cloud. Yeah, you
Matt Kimball 32:05
know, it's funny, because on the composability, side, and kind of the, you know, the goodness of what you're doing with a modular approach, I mean, even with, it's not even necessarily just PCIe versus, you know, something else, even within PCIe, right, a PCIe. Three was around forever, before it came out. And they're already they're already talking about five minutes, like customers that want to take advantage of these technologies, you know, it's almost impossible as an IT person to, you know, to within budget to be able to take advantage of, you know, these latest innovations in an easy way. And the other thing I would say, and I think you wonder sell Cisco on this is, you know, in a world where we as IT folks think hardware is somewhat commoditized, and servers are commoditized. It's because there's such a focus on, you know, everything being done out of software, right. And there's such an emphasis on the cloud, and it makes us kind of devalue the hardware. But it's really not that it's devalued. It's that companies like Cisco, are innovating in ways that, you know, it's making a lot of the complexities invisible to it, you know,
Unknown Speaker 33:11
we're trying to sup we are assiduously working to sublimate that for customers. Because you guys, you're right. I mean, and I appreciate you saying that, because I maybe I do, undersell it. Maybe it's Stockholm syndrome of a server guide and network. But, you know, we, we, we were trying to make it simple. And so we don't put the focus of our storytelling there anymore. Like we did maybe, as an industry 10 or 15. Year? Yeah. Yeah. I love it.
Matt Kimball 33:41
There really is, and it's, you know, I don't want to I don't want to call it as simple as plug and play. But you know, you you really do, you've done a great job of abstracting all of the complexities or, you know, a lot of the complexities from it and allow them, I think, allow IT organizations to be far more agile, and, you know, as he kind of talked about earlier, kind of be the consultant to the business, instead of running around plugging in cables, and, you know, patching servers, right,
Unknown Speaker 34:09
but friends, don't let friends engineer their own platforms, unless they have to. Right. Like, yeah, you know, so we're trying to we're trying to pre engineer this, pray and pray pre engineer where we can and, and help help people out.
Matt Kimball 34:23
I love it. Hey, can I ask you won't and Steve, I know we're running long. Can I ask you one more question though, Todd before?
Unknown Speaker 34:28
I'm yours. All right. Talk to us
Matt Kimball 34:31
about edge and Cisco's view of the edge and what you've been doing. I know. You know, you did some stuff with HyperFlex Express, you know, that has some edge applicability to it to can you hit on that in
Unknown Speaker 34:46
edge edge is exciting, right. And it's really again, it's like this just incredible spectrum of new experiences that we see customers trying to build. And we see it I think it's consumers. In our in our daily daily life, right in retail, in sporting, like in if you go to a modern day arena, like SoFi Stadium, where they just did the Super Bowl. That's, that's a customer of ours and, and I could sort of think I, I classify that as an edge environment because when I think of edge, I'm trying to think of, hey, where are there new sources of computing demand being created by new experiences, that to me is the definition of it outside the confines of either you know, traditional data center, or what cloud providers are providing as a proxy for the traditional data center. So something like sofa stadiums are, to me an edge case, and they're doing incredible things with the fan experience, right, and LA Dodgers, another customer for us, where they're trying to do all these incredible things for the fan experience, you know, starting with just, you know, hey, do I have good Wi Fi as a fan sitting in the stands, right? So that seems pretty basic, but things like, hey, which which concession stand as the shortest line for me to get my my ballpark state, you know, get, you know, so things like that right there. So wherever we see new sources, with computing demand emerging, it's really all over the map, like how heavy or how like customers need to go from a computing spectrum. In that environment. Some, in some cases, Edge means industrial, IoT, right. And we have products oriented on that segment, for example. But in with HyperFlex, we're kind of hitting that area where you need a little bit of horsepower, right? So say, a stadium, right, or a retail environment where you're doing video analytics, right of traffic, foot traffic in a in a big department store, and they're analyzing foot traffic. And you need a little bit of horsepower beyond what you would get in an industrial device or in a router, right? You need some horsepower. So with HyperFlex, you know, we've been working there to just lower the cost point, because obviously, you know, the stadium's not a good example. But maybe in retail, where you have hundreds or maybe even 1000s of sites, the cost per site is a big deal. And we need to skinny down the cost of that compute. In that remote environment, we need to tether it back to the mothership, you know, through something like inter site so that all those sites can be managed together. But what we put out there in that remote site has to be as light as possible. So we've taken hyperconverged clusters down to as little as two nodes. Yeah. So that we can, we can put those out in remote sites, and we've done some cool things. You know, when you when you get down to a two node cluster, your availability can be scary, right? So we provide like a third, yeah, there's there's some technology wonky stuff in here about, you know, witnesses and, you know, split brain if you lose connectivity to your cluster, and I won't bore your audience with but we use the cloud or not, we packaged up that witness functionality into a little container that you can run on, you know, some other device in that remote site like switch or a router. But our overwhelming focus, there's twofold, reduce the the cost and the footprint of the computing solution in those edge environments. And we've got a hyper flex edge, you know, optimized ruggedized, in some cases, for customers that need that offer, that gets down as skinny as two nodes, and then we're managing it all remotely from the cloud. But that is a really fun space, because that's where you see customers really coming up with the most innovative new ideas of how they're how they're changing the experience and digitizing it. And sometimes, you can't backhaul that and do the compute back in a data center in the cloud, you got to do it right there. And so just a tremendous variety of use cases. And that's really, really a fun space.
Matt Kimball 38:42
Maybe maybe in the next couple of weeks, we'll bring you back and do a deeper dive on on kind of the hyper flex space. And you know, a little bit more on on kind of what you've done, because you have a good story there that you're gonna I don't think our audience has heard as well as much. We'd love to love to kind of, maybe double click on that on the next one.
Unknown Speaker 39:04
I'm at your service. Well, I'll tell you what, let's let's collect some metrics on how this one did. And then, and then let's decide if you want me back.
Matt Kimball 39:17
Trust me, we'll want you back. And I've got you on tapes, and you're on our service. So that's awesome. Hey, Todd, thank you so much. Yeah, this was super helpful. And I love this kind of having these kinds of conversations where it's really more about customer experiences and what customers are, are facing these are far more interesting for our listeners. So thank you for
Unknown Speaker 39:38
that. Absolutely. Thank you, Matt. Thank you, Steven. Pleasure to be here.
Transcribed by https://otter.ai
Welcome to another data centric Podcast. I'm Steve McDowell and I'm here with Matt Kimball. Hey, Matt. Hey, Steve,
Matt Kimball 0:08
the Silicon Valley edition.
Steve McDowell 0:12
So we are road tripping in Silicon Valley. But our guests are not our guests are in Austin. So tell me, Matt, what are we talking about this week?
Matt Kimball 0:20
So what we're talking about. So, you know, we talk to a lot of customers about their journey to the cloud, the challenges they have with the edge and modernization in general. You know, how you how you deploy, provision and manage infrastructure and manage your environments, kind of holistically, and how much of a pain in the neck that can be right. Deal with it every day. And so, you know, we've we've talked to quite a few customers, and, you know, kind of some of the conversations that I've had, with a lot of the, the solutions folks in this in this space have been interesting, but one of the more has been with, with Cisco. You know, it's funny, because you and I, and we think about Cisco, and a lot of folks in it, think about Cisco and immediately go to their heads go to networking, or it goes to UCS, their server solutions that are that are quite powerful and quite popular. But there's so much more to what Cisco is doing right around management within our site and around HCI. So we wanted to bring in someone who knows everything about anything. A rush fan, a Cisco senior director, Todd Brandon, he's thots. The is the Senior Director of cloud platforms. He does a lot of product marketing, positioning, but he's been in the space for a while. He kind of gets all the dynamics with with the customer space. So that'd be great to have him coming in. Have a conversation with
Steve McDowell 1:53
us. You're talking about him like he's not in the room. Hey, Todd.
Todd Brannon 1:57
Gentlemen, thank you. That's some that's some high billing about everything. But then on my LinkedIn profile,
Matt Kimball 2:06
I'll probably do as a reference for you. How's it going?
Unknown Speaker 2:09
Never better. It's a beautiful day here in Austin. We need to do this from ski shores next time.
Matt Kimball 2:14
Oh, no doubt, no doubt. We'll actually be down there in the next couple of weeks back and back down there next couple of weeks. But yeah, Hey, Todd. So it's funny because Steve and I were talking last night for all the conversations we've had with you. And with Cisco in general. I think this might be the first time we've had Cisco on the podcast, and we're terribly sorry for that.
Unknown Speaker 2:37
So happy to help correct it.
Matt Kimball 2:40
Well, let's make this a running conversation. Hey, so But before before we get into like the meat of this, give it tell us a little bit about yourself, Todd, kind of where you know, what you've been doing, where you come from, you know, Cisco has been like,
Unknown Speaker 2:54
you know, I saw I started life in the army actually a long time ago. And I was an armor officer, so a tanker. So I had a lot of fun. Just leading troops was an incredible experience. And then when I finished up there I was looking for for a job. And I was here in Austin, and Dell hired, Dell hired the smartest 3000 People in Austin and hired about 30,000 more. I was I was in that best sucking sound going into Dell back in around 2000, or crushing 9495. I started there. And that's how I got introduced to servers and storage and workstations and kind of grew up a first job there was running benchmarks and found my way into product marketing. And then just kind of, you know, Battlefield promotions as the business grew type of thing. And in 2009, I was working in a group there that designed custom platforms for the cloud providers. So before they really went ODM director building their own, they were still buying servers from the likes of Dell and HP and others. And we had a really fun little business that did custom designs for, you know, the folks at Azure and Facebook and all the rest and but while I was working there, Cisco decided to get into the server business. And I was tasked to do competitive analysis on what they were talking about doing. I said, Well, that looks pretty cool. And so a few of us from Dell came to Cisco and then you know, folks resign and HP and there was a tranche of us that all joined Cisco back in oh nine to help the company get into the computing market. And I've been here ever since it's just been a real blast. I love the company, fantastic leadership and I'm really proud of what we do around it.
Matt Kimball 4:39
You know, it's so neat. It's funny that you went from kind of competitive analysis to fan to employee that's a green
Unknown Speaker 4:50
light with so much I bought the badge
Matt Kimball 4:55
Hey, so you know, for our audience, you know, obviously, you know, we speak to it, folks. You know, administrators, directors kind of up and down the food chain. And as I started in the intro, when you were in the green room, you probably heard us talking. You know, there's a, you know, there are a lot of challenges that it is facing today. Right? And it's not, you know, there's the promise of the cloud and the reality of the cloud, there is the promise of digital, digital digitization of your business. And then there's the pains of going of trying to get on the other side of that transformation, right? Is this what you hear it at Cisco, as well, when you speak with your customers and get feedback?
Unknown Speaker 5:35
Absolutely. I mean, the spectrum of adoption is as varied as you know, all the different customers and industries that they're in, right, and you've got, you've got kind of two minorities, I think, at either end of that spectrum, you know, you've got the folks that are like, Hey, I'm running on prem, and I'm happy as a clam. Don't talk to me about cloud cloud, they're increasingly a smaller, smaller minority. And then you've got folks at the other end that were either born in the cloud, or quickly got there and are 100% cloud native. But the vast majority are of the customers we work with are in that messy middle. Where they're working with usually, what we see is they're working with one primary cloud provider to try and get to some centralized kind of operational consistency. But inevitably, they end up working with multiple providers. And it's, it's nobody has a strategy to go multicloud out of the gate, maybe someday, right? Where they say, Hey, I want to have multiple suppliers to try and manage costs. But that really doesn't pan out, because most of the cloud providers give you increasing discounts based on how much of your footprint you allocate to them. Right. So the cost piece doesn't often work. But there's different reasons why people might have strategically, you know, adopted or, you know, entered into operations with multiple cloud providers. But most of the time, it's an accidental journey, right? Where there's like, Hey, we got to move into a new geographic region, or we have a, you know, very specialized requirement that takes us to a particular service on a particular cloud, you know, so most customers are in that situation where they found themselves in a multi cloud hybrid cloud operating model out of necessity, but they never really built the strategy for how they're going to operationalize that. And so yeah, that's, that's what we're seeing just kind of a really broad range of most people in that really difficult position and trying to figure out operationalize
Matt Kimball 7:30
it. So how do you as a, you know, as a, as a, you know, I'll refer to Cisco as an IT solutions provider, because you do more than, obviously, I mean, you run the full gamut from networking to, you know, to compute all the way to all of the software you do on the management side? How do you? How do you approach a market where there is no, I mean, every every customer is a unicorn in a way, right? Or a snowflake in a way? And that in their journey to the cloud and their adoption? And how do you how do you kind of come up and say the, this is how we serve the needs of the market in general? Well, I
Unknown Speaker 8:07
mean, I'll be the first to tell you, like, you know, when when people think of cloud technology, companies, we're not one, ironically, we're not one of the first ones that comes to mind often, right, the AWS so maybe, you know, the big tooling, like se hace corporate TerraForm are some of the first things that come to mind. But ultimately, you can't have cloud without networking. And when it comes to networking, we know a thing or two, because we've seen a thing or two, here at Cisco. And, and so we really approached each customer, kind of the way you described right, understand, where are they? And what are they trying to accomplish? Are they are they trying to, you know, expand or develop, you know, into new new markets? Are they trying to expand into new geographies? Or they're, you know, they trying to adopt? Is their strategy driven more around an operational set of questions like, hey, we feel like, we shouldn't be in the business of managing our own data centers, we want to get this to the cloud. So customers come at it from a bunch of different angles. And like you said, we have a technology footprint that is so broad that from a storytelling perspective, which is a lot of what I enjoy, you know, trying to figure out here at Cisco, it almost collapses under its own weight, because we have technology that ranges everywhere, from connecting users and devices, to data and applications on the other side and on the cloud, we and everything in between. And so it's really that that's where our sell our sellers and our our, our solution partners in the channel, really earn their their money, right in that in that consultative approach and figure out is what we don't want to do is walk up to customers with a kind of one size fits all strategy. And when and when I see you know, the the industry's kind of presented customers with two basic choices right? In terms of how they approach it, it's like, Hey, we're either going to help you lift and shift your on prem operating model to the cloud. And that might be most exemplified by say, VMware his approach to the problem. Or you've got it from the cloud provider side where they say, Hey, we're going to help you bring the public cloud into your dataset with a solution, like outposts, for example. And those have a lot of merit. But they they're also a little bit inelegant, in a way. Yeah. So everything that our engineering teams are doing here at Cisco is really kind of focused on how can we, how can we create operational integrations that work at a more fundamental altitude, right. And maybe an example might illustrate this, right when, when we think about how to do cloud networking, for example, one of the technologies we have is our ACI framework around data center and cloud switching. And rather than take, you know, an existing environment of ACI try and just plop it in the cloud, what we did is we did API level integrations with all the major cloud platforms so that you take your policy that you've developed, and you can translate it into the the, you know, the very atomic knobs and dials that you would turn, so that you're not imposing a new networking schema, and Amazon or an Azure or Google, rather, we're doing kind of a universal translator, where we go out and use the API to dial in the networking paradigm, you know, within that cloud provider to get to the end state you want. Yeah, without you having to necessarily learn how they do. Because, you know, all the providers have a slightly different approach to networking. And, and rather than have to learn all those, and rinse and repeat, we act like a universal translator, almost for policy. And so it's at a more fundamental primitives kind of level, right? That we try and that we try and do our engineering, and with a very cloud agnostic approach, because we got customers are all over the gamut of who they want to work with. And so we don't really want to take, we don't we don't want to, you know, have a favorite, if you will, in terms of which cloud providers we're, we're engineering to it's, it's across try be very agnostic there. So I think that's what's a little bit different about us is we're trying to work at a much more down in the primitives, and kind of help customers operationalize in a more, yeah, maybe cloud natives or a highly charged word, but in a more cloud native way, right?
Matt Kimball 12:23
Yeah. Well, I think I think you bring up a good point there too, in it from a little bit earlier on, and that, you know, when when you talk about, you know, their their, their, their extremes on on either end of kind of the cloud adoption continuum, if you will, most people live in that messy middle. And amongst most companies, and, you know, the in the ability, the ability, or the the capacity for IT organizations to fully optimize, you know, their on prem to a multi cloud environment is next to impossible, right, you don't have enough people, you don't have enough of the right skill set, you don't have the right technologies, and to what you said a lot of a lot of IT solution providers, you know, say here is our solution, make it fit in your, you know, and, and they try and build something that suits the needs of everybody, maybe not quite as well, whereas it feels like Cisco is certainly more more understanding of all of the phase of issues facing in, you know, like Tower of Babel, right, that allows me to talk to any cloud in a very in a much easier fashion. But when you get to the operational side, it's one of the areas I want to jump into, because we've talked to a lot of companies about this, the operational side and managing. You know, one of the things I was impressed with, and I'm always impressed with when I walk away from our conversations, is what Cisco has done with inner sight, which I don't think gets an nearly enough attention in the market. Can you kind of, you know, for our our listeners, who may be, you know, aren't familiar within our site, can you give a 10,000 foot level?
Unknown Speaker 14:03
I appreciate you bringing that in? We, you know, we the reason I move to Cisco was how I saw the the management paradigm being developed around compute. And, and at that time, you know, things were very Device Centric at that moment in time. And so you had management modules and blade chassis 's and you had you had all these different management elements that would be spread across the data center in what Cisco was trying to do is consolidate those in the network fabric. And in that threw off a lot of benefits Policy model for all that. And so that was where we got started with Unified Computing System. And so we drove down that road in about 2015 2016, we realized we need to take this control plane from the network fabric to the cloud. And so that's how intersite really came to be we you know, we launched it in 17. And initially The, you know, the the value that we were providing with intersite was related to infrastructure on prem. But the value prop of moving it to the cloud was just all the benefits of a SAS approach. So, hey, we can connect your operators anywhere with your infrastructure anywhere. So Riot Games, right? They do Liga, legends and esport. Gaming, right? They've got 12, I want to say 12 broadcast studios around the world, where they bring in gamers and do tournaments and broadcast it to the world. And in a scenario like that, you know, you don't want to have to have hands or talent in each of those eSports facilities that know how to configure the servers that the games are running on. So right, so So riot came to us and, and we suited him up with nexus for ultra low latency switching in these eSports arenas, because that's obviously, you know, really important to gamers and their tournaments. And we have UCS compute. But then within our site, they were able to look at their all 12 of their Esports arenas around the world. And that all that compute in those in those arenas. And from an administrative perspective, it looks like it's in one datacenter. And so if they want to do something like hey, and this is really tactical, but it kind of shows the day to day, right, if they want to disable cores on a server, so they can overclock the other cores, and soup up the gaming experience. Rather than go in and have local hands doing that machine by machine around the world, they can set that very granular policy element into the definition of what it means to be illegal legends server, whatever their game files, their their gaming, and then they can they can apply that uniformly to all their machines around the world. They're doing it from the cloud. So our value prop with intersite was originally Hey, what is SAS managing infrastructure as SAS from the cloud? How does that benefit you? The other thing it did was, it got us out of a mode of chasing our customers around saying, Hey, have you upgraded to the latest version of UCS Manager, right, etc, etc. Now, they're just logging into intersite. And any new features we deliver just show up. And they don't have to think about building and maintaining a monster to manage their monster. So that's where we got started with with UCS. And we did some things like connected tax. So we're kind of in the data center with our customers where they want us, and we can alert them to problems and we automate, you know, problems that we identify, we give an automated remediation. So a lot of things that we're doing are just trying to take, you know, day to day administrative burden off of customers. And so that was the maniacal focus initially. And then we realized, okay, you know, infrastructure today no longer means hardware on prem, it means the abstracted layers, right? Virtual machines and containers. And so we started to build, particularly kind of early last year, we started building out services on top of this platform that kind of address that. And so we started with some things around containers, we have container service, we did some workload optimization, so you can decide where to place workloads on prem or in the cloud, and lower your closet, there's like three or 4 million SKUs instance types, right? Across all the cloud providers. How do you keep up with that? So we have an optimization service that is keeping up with that and says, Hey, you're on instance, type A, based on your workload profile, we think you should be on instance, type B, would you like us to do that and save you a buck or two? So we've kind of been moving up the stack now, out of the on prem infra, into these elements that span hybrid cloud.
Matt Kimball 18:39
I mean, that's, that's killer, right? Because it for a number of reasons, right? First, first, is it to me, you know, when you think about the early days of intersite? To me, that's the realization of what, you know, large enterprises used to try to achieve through their Knox, right? I mean, and there was always that notion of single pane of glass, but it never really kind of was realized in a real and simple way. But as you move up that stack, you know, it sounds like you're almost talking about managing all of your cloud operations and, and acting as somewhat of a broker, or at least a suggested broker to help, you know, not just operationally but economically as well, for these IT organizations.
Unknown Speaker 19:21
Yeah, that to an extent, definitely. You know, and I would, you know, I, I really shy away from trying to portray intersite, or any platform really as the one platform to rule them all because I was with a, I can't name the company, but I was lucky enough to be in a conversation with a CIO and some of our executives and, and she said, Look, this is how we basically operate it today. That is, the business comes to us looking to create a new experience, right, whatever that experience may be. We go off and develop a solution that answers that need to deliver that experience. But then the next step for us is to look outward. In the industry of platforms that we can then integrate to create that solution, right? They're not building it from scratch. So that that's what's driving that, that that paradigm, right or that that approach that the IT teams are using today, is really driving our engineering and innovation agenda. And in the sense that we're taking all the technology at Cisco, that used to be in a lot of disparate systems, or embedded in products. You know, step one is let's salsify, those things and move them to the to a cloud platform. And then let's aggregate them into logical buckets, based on, you know, the users that are going to use them because again, you may not want to aggregate all the stuff you've got, you just want to kind of get it into the right chunks for the operators that care about security or net ops or, you know, computing infrastructure. And so our platform strategy at Cisco is really aimed at how can we aggregate this technology into easy to use platforms, and then create integrations between them that help our customers, you know, bring us all together. So integrations between platforms that share telemetry data, for example. So you've got two operators may be using App dynamics, for example, to look at the application topology, you know, we acquired AP, the A few years back, fantastic visualization of the application environment, well, we integrated that data into intersite. So that not only can the users using App D, see how applications are connected down into infrastructure. But now the infrastructure cats using intersite can see how the infrastructure that they're accountable for, for operating is tied to the application and all the distributed components of that application. So those kinds of integrations, help our customers get to, you know, a single source of truth between teams. So, you know, instead of feeding revenue to the local pizza shop, by having all night, warm sessions to root cause problems, you know, it's on us, we think, at Cisco, to bring a lot of this telemetry data together behind the scenes, to give a rich experience to different platforms, but have it all feeding off a single source of truth, a common language. So it's very easy that that's the biggest challenge people have is just seeing how everything's connected. And that's what we're trying to discuss how can we help customers connect the dots and give them situational awareness? That's what we refer, you know, full stack observability, right? It's a huge topic in the industry, kind of moving beyond monitoring. But really understanding how the pieces are connected in real time. So you can get ahead of problems. And that's really where our focus is, is this platform strategy and the integrations between them. So we can help those CIOs stitch things together.
Matt Kimball 22:43
What are your views? And I know this is a little bit off topic or not really, but what are your views on kind of autonomous, you know, heading down the autonomous infrastructure route me because I think about what you're saying, right? And, you know, you have you do have the tool chain has support full stack observability. And he talked about AppDynamics, intersite, and all of the other pieces in your portfolio. And it seems like you know, that the next natural extension is not just hey, you can see what's going on. But these problems are root caused and resolved before they even happen, right. Predictive failures or, you know, whole story. Yeah. What are your thoughts on that?
Unknown Speaker 23:25
I'm smart. I know, we're an audio podcast here. I'm smiling. Because, you know, I T operators are paranoid, unnecessarily paranoid and conservative. Bunch, right? I mean, they're the ones who get lit on fire when stuff goes wrong. So the idea of, you know, you know, autonomic operations is been around for a while, and there's definitely, you know, ways that you can, you can do that. But every customer I talked to generally says, look, keep me out of trouble by alerting me to things, right. And then give me the easy button to fix it. Yeah, but I want to be in the middle. Right,
Matt Kimball 24:05
I want to press that button,
Unknown Speaker 24:06
I want to I want to be alerted. I want to press the button right now. Now. Now. Now, what's what's starting to evolve though, what I what we see starting to happen, though, is maybe low stakes decisions. They're willing to fully automate. So an example might be that instance type scenario, where, hey, you know, instance, a is going to save you a couple bucks over instance, B, they're both on AWS. You know, you're not going to lose, you know, nothing's really you know, there's nothing structurally changing in your in your overall topology. So maybe those are, you know, things that I'd be willing to flip, flip the full auto, if you will, but I think it's graduated in that way. So low, low risk, low risk elements. Yeah, some people are starting to adopt the full automation strategy before. I think the mainstream IT teams, they're still in the mode of I want to be in that decision chain, but many Make it easy for me on both sides, right? You know, you know, help me separate the signal from the noise with with good analytics, help me connect the dots, and then give me the easy button to fix it. And that's what I see there. It's very, very, I haven't met a customer that is like, hey, put me on full auto forever that
Matt Kimball 25:18
oh, yeah, no, it's funny, because it's, it's a story we all like to tell. Right? And it's, I mean, as an analyst who, you know, we can say whatever we want not be held accountable. Even though we haven't
Unknown Speaker 25:30
marketing, it's the same thing.
Matt Kimball 25:34
Very, very thin line between you and I, my friend. But you know, it's a, it but you know, it's easy to kind of build these grand visions. But you're right. I mean, concern it and it has a right to be conservative and their adoption of technology. And it is certainly a slow moving train. You know, what's funny is we are we are 30 minutes into this conversation. And we haven't we just got started. I know. And we're on with Cisco, and we haven't talked about hardware at all.
Unknown Speaker 26:05
Oh, man. I know, there's some some server augers out there that one of your slots and watts will have to come back?
Matt Kimball 26:11
Well, but But before we can, before we drop, I want to kind of give us a little bit into you know, it. It's funny, I say that about, you know, we were on for 30 minutes. And it's been a software discussion. But, you know, Cisco is a is a is an evolving company, in my mind from kind of where they were even just a few years ago. And it seems like, you know, you're one of the few companies that seems to understand this evolution of it, and the the weird journey that it's taking to the cloud, right, that messy metal and trying to meet folks, but, you know, you do have a good hardware portfolio. And there are a lot of different needs out there with with edge with your traditional compute with, you know, AI and the higher end workloads. You know, I know you've done some stuff around HyperFlex and UCs X, do you want to
Unknown Speaker 27:03
know? Yeah, absolutely. Right, I think that's what kind of sets us apart in some ways, is we're innovating up and down the stack, from silicon, all the way up through the things we've been talking about in the operational model. And where we still see opportunities to deliver value through silicon innovation, we're still doing it, we just released our latest incarnation of our virtual interface card, right? That's a custom ASICs that we build. And it's a converged adapter that you can, you know, you know, it can transform itself to rep, you know, appear to the operating system and appear to the machine that it's hosted in, as you know, 256 virtual adapters of any kind, you want to build Ethernet, or HBA. And to do that, you've got to develop some custom silicon to deliver that kind of capability.
Unknown Speaker 27:53
The other thing that it kind of dialed in on computing, the computing architectures have really changed dramatically, just in the last five years, right, it was kind of quiet there for maybe a decade where you had your classic food groups have, you know, processor, memory storage, but now storage is coming a lot closer to compute with NVMe. The, you know, AMD and Intel have done some incredible things with core density and options there. So, you know, the architecture, how you build in fabrics, right? Cx L, you know, you know, PCI, and where that how that's evolving the fabrics and how the subsystems of a server are interconnected. There's a lot of different ways to, to approach it today. And so we, you know, we've done some things there with this platform introduced last summer called X series, and we kind of looked at the problem, I won't say, you know, composability, right has been this, this, you know, this this vision or Mirage, I think, on the horizon for a long time in the computing, architecture circles, but there's some things starting to happen where we can start to disaggregate components and then assemble them, you know, the way we want, you know, from, you know, through a software control point. And we're doing that with our X series product. You know, in our announcement this month, we, we introduced basically a PCIe blade, right? So you've got server server nodes in that in that X series chassis. Now we have a node that plugs in that you just populate with PCIe devices, like GPUs, for example. And so we've introduced this architecture, we have a, you got a classic converged fabric that I kind of talked about, you know, for Ethernet, San management traffic, but we added a second fabric. It's really we call it x rabbit, but it's it's for the purpose of connecting server subsystems together. So now we can take a workload that you would typically have gone out and maybe had a specialized system for maybe you had an external array like that, you know, there are vendors that have the GPS You know, boxes basically, and then you know, cabling hooked up to your servers, we've kind of streamline that with our x fabric so that you can assign resources from that GPU blade out to that server nodes and, and do that through policy and kind of assign the GPU, which is a very expensive resource relative to others in the system, you know, assign that to the nodes that need it when they need it. And so, so yeah, there's a lot going on in server architecture. And we're having some fun with it. Because we're all about fabrics. We're all behind connect stuff. And we got buildings, real engineers, are good at that. And so our server architecture is still over the horizon. I mean, we think we made some big waves when we introduce UCS, back in Ohio, and now we built built our, our chassis Foundation, our architectural foundation deck series for the next, whatever, 10 or 12 years and, and by, and one thing we're trying to do is decouple all these subsystems, because they all they all kind of innovate on different cycles, right? You get you get innovation from different components in the industry on different cycles. So we want to make it all modular so that, hey, in two, three years, like today, our fabric is PCIe based, maybe we want to use a different fabric and two or three years, it's just, you know, swap out those x fabric modules. And now you're on a different fabric for the interconnect for all the subsystems within within the chassis. So yeah, we're having a lot of fun with with baseline architecture at the product altitude. But then the magic is how do you connect that up into the operating environment that we do with, in our case, intersite. So that you're not creating more complexity? That's the key, right? Like the big problem. composability has been, you know, who wants to have to deal with the complexity of how to put all this stuff together? How do we make it automatic transmission versus a manual? Right? And so, how you tie those physical architectural decisions and opportunities? How do you make that easy for customers to consume the benefits of those things? And that's where we focus with a control point within our site and doing that from the cloud. Yeah, you
Matt Kimball 32:05
know, it's funny, because on the composability, side, and kind of the, you know, the goodness of what you're doing with a modular approach, I mean, even with, it's not even necessarily just PCIe versus, you know, something else, even within PCIe, right, a PCIe. Three was around forever, before it came out. And they're already they're already talking about five minutes, like customers that want to take advantage of these technologies, you know, it's almost impossible as an IT person to, you know, to within budget to be able to take advantage of, you know, these latest innovations in an easy way. And the other thing I would say, and I think you wonder sell Cisco on this is, you know, in a world where we as IT folks think hardware is somewhat commoditized, and servers are commoditized. It's because there's such a focus on, you know, everything being done out of software, right. And there's such an emphasis on the cloud, and it makes us kind of devalue the hardware. But it's really not that it's devalued. It's that companies like Cisco, are innovating in ways that, you know, it's making a lot of the complexities invisible to it, you know,
Unknown Speaker 33:11
we're trying to sup we are assiduously working to sublimate that for customers. Because you guys, you're right. I mean, and I appreciate you saying that, because I maybe I do, undersell it. Maybe it's Stockholm syndrome of a server guide and network. But, you know, we, we, we were trying to make it simple. And so we don't put the focus of our storytelling there anymore. Like we did maybe, as an industry 10 or 15. Year? Yeah. Yeah. I love it.
Matt Kimball 33:41
There really is, and it's, you know, I don't want to I don't want to call it as simple as plug and play. But you know, you you really do, you've done a great job of abstracting all of the complexities or, you know, a lot of the complexities from it and allow them, I think, allow IT organizations to be far more agile, and, you know, as he kind of talked about earlier, kind of be the consultant to the business, instead of running around plugging in cables, and, you know, patching servers, right,
Unknown Speaker 34:09
but friends, don't let friends engineer their own platforms, unless they have to. Right. Like, yeah, you know, so we're trying to we're trying to pre engineer this, pray and pray pre engineer where we can and, and help help people out.
Matt Kimball 34:23
I love it. Hey, can I ask you won't and Steve, I know we're running long. Can I ask you one more question though, Todd before?
Unknown Speaker 34:28
I'm yours. All right. Talk to us
Matt Kimball 34:31
about edge and Cisco's view of the edge and what you've been doing. I know. You know, you did some stuff with HyperFlex Express, you know, that has some edge applicability to it to can you hit on that in
Unknown Speaker 34:46
edge edge is exciting, right. And it's really again, it's like this just incredible spectrum of new experiences that we see customers trying to build. And we see it I think it's consumers. In our in our daily daily life, right in retail, in sporting, like in if you go to a modern day arena, like SoFi Stadium, where they just did the Super Bowl. That's, that's a customer of ours and, and I could sort of think I, I classify that as an edge environment because when I think of edge, I'm trying to think of, hey, where are there new sources of computing demand being created by new experiences, that to me is the definition of it outside the confines of either you know, traditional data center, or what cloud providers are providing as a proxy for the traditional data center. So something like sofa stadiums are, to me an edge case, and they're doing incredible things with the fan experience, right, and LA Dodgers, another customer for us, where they're trying to do all these incredible things for the fan experience, you know, starting with just, you know, hey, do I have good Wi Fi as a fan sitting in the stands, right? So that seems pretty basic, but things like, hey, which which concession stand as the shortest line for me to get my my ballpark state, you know, get, you know, so things like that right there. So wherever we see new sources, with computing demand emerging, it's really all over the map, like how heavy or how like customers need to go from a computing spectrum. In that environment. Some, in some cases, Edge means industrial, IoT, right. And we have products oriented on that segment, for example. But in with HyperFlex, we're kind of hitting that area where you need a little bit of horsepower, right? So say, a stadium, right, or a retail environment where you're doing video analytics, right of traffic, foot traffic in a in a big department store, and they're analyzing foot traffic. And you need a little bit of horsepower beyond what you would get in an industrial device or in a router, right? You need some horsepower. So with HyperFlex, you know, we've been working there to just lower the cost point, because obviously, you know, the stadium's not a good example. But maybe in retail, where you have hundreds or maybe even 1000s of sites, the cost per site is a big deal. And we need to skinny down the cost of that compute. In that remote environment, we need to tether it back to the mothership, you know, through something like inter site so that all those sites can be managed together. But what we put out there in that remote site has to be as light as possible. So we've taken hyperconverged clusters down to as little as two nodes. Yeah. So that we can, we can put those out in remote sites, and we've done some cool things. You know, when you when you get down to a two node cluster, your availability can be scary, right? So we provide like a third, yeah, there's there's some technology wonky stuff in here about, you know, witnesses and, you know, split brain if you lose connectivity to your cluster, and I won't bore your audience with but we use the cloud or not, we packaged up that witness functionality into a little container that you can run on, you know, some other device in that remote site like switch or a router. But our overwhelming focus, there's twofold, reduce the the cost and the footprint of the computing solution in those edge environments. And we've got a hyper flex edge, you know, optimized ruggedized, in some cases, for customers that need that offer, that gets down as skinny as two nodes, and then we're managing it all remotely from the cloud. But that is a really fun space, because that's where you see customers really coming up with the most innovative new ideas of how they're how they're changing the experience and digitizing it. And sometimes, you can't backhaul that and do the compute back in a data center in the cloud, you got to do it right there. And so just a tremendous variety of use cases. And that's really, really a fun space.
Matt Kimball 38:42
Maybe maybe in the next couple of weeks, we'll bring you back and do a deeper dive on on kind of the hyper flex space. And you know, a little bit more on on kind of what you've done, because you have a good story there that you're gonna I don't think our audience has heard as well as much. We'd love to love to kind of, maybe double click on that on the next one.
Unknown Speaker 39:04
I'm at your service. Well, I'll tell you what, let's let's collect some metrics on how this one did. And then, and then let's decide if you want me back.
Matt Kimball 39:17
Trust me, we'll want you back. And I've got you on tapes, and you're on our service. So that's awesome. Hey, Todd, thank you so much. Yeah, this was super helpful. And I love this kind of having these kinds of conversations where it's really more about customer experiences and what customers are, are facing these are far more interesting for our listeners. So thank you for
Unknown Speaker 39:38
that. Absolutely. Thank you, Matt. Thank you, Steven. Pleasure to be here.
Transcribed by https://otter.ai
