 Live from Barcelona, Spain. It's theCUBE, covering Cisco Live 2020. Brought to you by Cisco and its ecosystem partners. Hello everyone, welcome back to Cisco Live Barcelona 2020. Kick it off. The new year, of course, is theCUBE's coverage of four days of CUBE action. All day I'm John Furrier, my host Stu Miniman, the two great guests, Thomas Sheba, Vice President of Cisco and Yusuf Khan, Vice President of Technical Marketing. All things data center and networking, these are the guys. Guys, good to see you again, welcome back. Thanks. Always fun. So kicking off the show, I know that some announcements coming, so we're going to save the good stuff for tomorrow and Wednesday, but a lot of new things going on, data center and Cisco ecosystem. Give us the update. Yeah, again, thanks for having us on. So yeah, I mean, there's actually a lot of good stuff on the data center site. Let me touch a couple of items. One of, we kind of started two years ago, actually, was assurance. We're expanding our analytics portfolio, we're adding the insights capability. So it's the assurance and network insights tool set. Very, very cool stuff. Really focused on the network operator, and it was one of the messages we got. You guys need to help us here in these complex cloud environments. And so what we have is we built ACI extensions for our fabric controllers, Balsa, Nexus and ACI site. Same thing, pure software extension, and initial feedback from customers is very, very happy what they see. So that's one piece. I don't know, Yusuf, you want to take a little bit on what we do with ecosystem partners? No, thank you. Yes, I mean, we are very excited also to announce some of the new integrations that we have with our ecosystem partners. And for example, Elgosec and ACI integration, Terraform from HashiCorp and ACI integration, continued expansion with our Splunk apps with the ecosystem. So these are some of the new things that we are working on. So that is excellent. And on top of it, Thomas, you can expand on it, but I think we are very happy that our 400 gig portfolio is shipping now, and we have customers in production on our 400 gig portfolio. So that is great news for us. Yeah, that's actually a good point. You mentioned Splunk and Terraform, HashiCorp, you know, ecosystem partners. It's interesting, if you look at the performance of a lot of those companies, cloud was as a tailwind for them. So because the consumption is a service, the customers are all embracing it, but it's not just public cloud. It's the data center now is back. So can you guys just share your thoughts on your environment with your customers because the software is the key, get it as a subscription or consumable model. What are some of the trends with the consumer, I mean, customers in the data center because cloud and hybrid now is happening and it's real growth. Oh, it's absolutely happening. So yeah, I mean, maybe a little bit of why this is happening while we're having some of these integration. You're absolutely right, cloud is happening, but really cloud means hybrid cloud or for some customers, multi-cloud hybrid because they're going to have two different cloud providers. But it's really a hybrid cloud. So it's really distributed data center. And so the interesting piece happens, it's really two things that need to come together. There's this whole network automation analytics, which is how do I get from my data center into a cloud and how do I treat this really like the utility? But that's the infrastructure, but then there's this front end because what really drives this is the application refactoring. And this is where the application automation needs to come together with the infrastructure automation. And so that's one of the reasons why we have this integration with Terraform and the other one is like a Jenkins pipeline tool. How do we actually take what the application was doing in front end and then seamlessly make this into infrastructure, which is like, as you probably know, the infrastructure as a code sync. And that doesn't really matter whether that's in the cloud or on-prem that has to work across. That automation is a huge thing. And it's so nice to hear because Thomas actually, when Cisco first came out with application centric infrastructure, I kind of looked at a little bit. I'm like, well, come on, how much you're actually tying to the application? Well, it was Cisco skating to where the puck was going. And I think that the technology today and what you're talking about is closer to that application. We have, you know, we're here in the DevNet zone. We're talking more about those pieces, not just, oh, it's something that runs over the pipes and I've got buffers and traditional networking pieces. So would you say that's fair that we're a little bit more application centric today in 2020 than we might have been a couple of years ago? It's actually, it's a very good comment. I'm probably would spin it slightly different because I'm the pragmatic guy. Yeah, do we want everything at the same time? Absolutely, right? But you do have to put some of the building blocks in place. And yes, application centric really meant more, we changed the configuration management scheme of infrastructure from thinking about network terms to using application terms. And that's really what application centric means. It doesn't mean you change your application. It was more like change the paradigm, how do you manage infrastructure to not just automate, everybody does that, but actually have an abstraction layer that is meaningful to secure an app to people. And you're right, it takes time to get there. In the end, customers and users are looking to deploy applications faster, manage applications better. That's the whole purpose of building the data center so that we can host the application. So what we did is we introduced constructs that can help you manage those applications better, deploy them faster, manage the life cycle of those applications faster. And that's why we introduced the constructs. And again, going back to your comment in terms of buffers and switches, we firmly believe that the plumbing, which is the networking, has to be state of art for us to abstract these things on top through software and exploit through software. So we have to have a best in class network and the switches, and then we have to have build the abstraction that we can exploit through the software means. And also that highlights the partnerships that you mentioned, companies like Splunk and HashiCorp, they're living in a multi-cloud environment. So I shouldn't need to think about for some of them, oh wait, is it hybrid cloud, public cloud A or my data center, things like that? I'm going to have that common tooling and skill set across those environments. Because all the CIOs that we talked to, I mean, multi-cloud is part of, big part of their strategy. And they want to make sure that they have consistent security posture, whether it is on-prem, whether it is on multi-cloud, or consistent governance model across hybrid cloud. Yeah, that's a good point. I want to get your thoughts on that because multi-cloud and hybrid, which both mentioned is interesting. And what we were saying in our opening segment just earlier, multi-cloud is a business problem. It's what you have. It's a situation. Yes. Hybrid is technology. You're implementing new things for an operating model that hits core to what happens in your environment, whether it's software development, application awareness, network automation. So they're two different things, but they're kind of related, right? You nail hybrid with public, private, or public on-premise, then then multi-cloud can be dealt with. This seems to be where you guys are fitting in, right? Because you can do the hybrid public, then you connect, just that's the outcome of the software. You're spot on, right? It's, people use it, and sometimes it means the same, and sometimes it's really not. And it's really, hybrid cloud is really around, how can I extend my data center to a public cloud infrastructure, right? And that's more of a technology discussion. What do I need to do to make that happen? And then there's the multi-cloud discussions really around, how do I have consistent policy? Because I want to get to a situation where I don't have to worry. And so I can deploy those, I can describe as when to employ wherever I want it. And so you're right, there are two distinct things that need to happen. But I do, sorry, I do want to come back to your comment because I can't have the energy there. But to use this as a comment there, right? I mean, for half these application developers, they want to use tools like Terraform or Jenkins or... Ansible. Ansible or Splunk, all of them expect that they have an API and they expect actually a network API. What they all prefer to have is something that makes sense from an application construct perspective. And so that's why I was like, we have to put something in place to make that work, right? Was it they want all there that the application team could jump? Clearly not, but it's very clear. I mean, if I look, we are now about six years into this. If I look back, I think it really jolted the market and I think it got everybody moving in that direction. Yeah, and again, when we use the term application centric infrastructure, the whole purpose is it is conducive to deploy applications faster and manage applications better. That's why, right? Yeah. Okay. Wonder if you can dig in a little bit on the 400 gig. Tell us, you know, it's not just the next step function. We're trying to go more to the applications. You talk about these changes. So what do people need to understand about 400 gig? You know, what's the same? What does this unlock for me? Does this tie in with all my Wi-Fi 6 and 5G and everything else that I'm doing? And, you know, where and when is this most important? Well, let me take it maybe on 400 gig. A, it is available in shipping. A little sneak preview, we're actually going to have a customer who visits on Wednesday talking about what they do with 400 gig. Awesome. European data center as a French customer. 400 gig is really, it's an evolution. I mean, the way I look at it is, right? I mean, we had one gig, 10 gig, 40, 100, 400, right? It's literally an evolution. And we're always looking back and saying, well, do you really need that much bandwidth? And, you know, you later, you know, when you ask that question, it's like you look like you missed it. Where's it deployed today? Service to water. No doubt around it. It's all in the service to water space. It's primarily what we call a large scale cloud providers, but also the initial more technical disease are looking at this as evolution. How do we build 400 gig? The way we approach this is, this is not something special. Everything that we do today around ACI, everything we do around analytics has to work, right? Because customers are not building around speeds. Customers building around the operational model and whatever they have has to work. Just because I've got my 4X speed, that has to work the same way. And so 400 gig for us is really an extension of what we have and you will see it, it plugs in directly. So can I build a 400 gig ACI fabric? Yes you can, if you want to. With all that horsepower, obviously the next logical question comes to my mind is, okay, faster means more data, that means more potential fat finger mistakes on configuring, but if you automate that away, you need AI, right? So analytics and AI become interesting to that. What's, how does that fit into the customer journey when they go, okay, I'm getting going faster, from application aware, is there an analytics angle on this? Yes there is. No, you're absolutely right. I think based on the survey that we received, I mean US corporations are spending billions of dollars due to the IT outages, right? And most of those outages are human errors, right? 43% of the IT corporations are like spending 43% of their time in troubleshooting those outages. So I think it is very, very important as the data center are scaling, as the fabrics are getting automated, is that we equip them and provide them with the operation tools that can look smartly and proactively predict the network changes. They can assure that intent, the business intent has been translated into the network and proactively tell them that what are the problems they might run into. And when they run into the problems, also intelligently explain it to them that what is the correlation of the events that they see on their log files and what is the root cause of the problem, right? Yeah, you get a lot of data to work with there and experience, right? That's where the predictive analytics are. That's where it is. We let me expand a little bit. So I started off with saying we have this interesting extension of network insights, which is precisely that what you said just a library on. It's really an engine that takes telemetry data and we're going actually one step further than everybody else that I know. Everybody talks telemetry, but they talk about like software telemetry, network state. We actually can marry that up with actual traffic data at real time and we can give you that correlation. And now I'm getting actually where you are can see I kind of go into is, I can actually tell you what's the root cause between why do I have a congestion? Why do I have a problem and who is impacting who causes this? And I can actually predict the stuff. I can actually see this before it happens and now help a customer. I can look at other customer experience and I do really more as a machine learning. There's really an opportunity there. We just we're just scratching the surface if you ask me. I mean, there's so much upside. I mean, historically speaking, if you look at it, I mean, we had all the show commands in the world which can tell you that what they basically just look like, what the cam utilization is, but the correlation or the time-based correlation was missing in terms of when you, you're seeing some traffic degradation. You don't know whether it is drop, drop on what switch, which type of traffic is getting affected. Now we have the ability to, and using ML and AI techniques to correlate these events and give you a meaningful picture back to the customer. So you can pinpoint that, look, my video traffic on switch number five is getting affected because there is a drop in the output buffer because my link is congested. And that only works if you have quality data. It's not so much volume. I mean, the faster you go, Facebook and these guys prove it, you need to use machine learning, but if the data is good, then the outcomes are better on the predictive. You need a half the flow data. If you don't have it, that's nothing you can do. So. So scale is something we talk a lot about in the network. When I walk through the show floor, I'm starting to see some of the small scale because we're talking about edge computing, talked about shrinking down some of the things we're doing. When I hear telemetry data and AI and everything, I'm like, oh, here's some big opportunities that we need to attack at the edge. So what can you tell us about where your group is with some of the edge pieces? Well, interesting. I actually just came out of the a service provider opening session and I was together with a T-System actually on stage who is a customer of ours. He's using actually an ACI Fabric together with a SIVAM environment, which was like a virtual infrastructure management of X86. And they're using that in a telco cloud environment and clearly as an interconnect for networking services and it's going to move if you look at what they have in mind, moving in motor edge services. And that's an SP example that we have today deployed. But clearly, I think you're going to see this in enterprises, you see this pretty much in every customer base, right? Because what you do have is you have to straight off between do I want to get all my data back centrally or do I want to compute on the edge? And what we have put in place was our ACI Fabric, I can run this in a highly distributed and still scalable environment, managed centrally with policy. So not only is this actually where we sync the word, it's going, we actually have customers doing this. Yeah, I think it's a tell sign too. And I think just my final question for you guys is, and we've been saying, I've been saying this in the queue with the team is cloud helps everybody if hybrid kicks in, which we now has proven that hybrid cloud is a reality, that's what's going on. Technically, operationally, if you believe that, then you go the next level, which is cloudification value. So I want to rattle off some key words for you guys and I want you to respond to them. So, cloudification of networking, network as a service, way into cloud versus internal, SD-WAN, simplification of the edge, BGP, security and networking, common policy. It's a lot of technology, good be good. That's all sounds complex, but it's got to be simplified. What's your reaction to that cloudification? How does that kind of direction package itself out for the benefit of customers? Because there's a lot in there, right? SD-WAN alone. There's a lot in there. So, I look at this, it's in the end, it's a business. It's that simple, right? And what's going on? You want to generate more revenue, more services, which is where the profit and the money comes from. And you have to scale, which means more service individually, more scale, how many customers you're going to deliver to, how fast you can roll this out, without having your costs going up the same way. And that's really in the end what it comes down to, at least in my book. And then you make your decisions what you're going to pick, right? How do I figure out how to develop an app faster? Maybe you're going to go to cloud, to start cloud first to develop, and then you figure out, oh, I need to hit a certain scale, I going to start having it running here and running here, my deaf ear, my production here, I need to connect it. But all of these things in the end coming down, how do you roll out services faster without my cost actually going up, but preferably saying flat or going down? So business model. It's a business problem, it's what it is. Yeah, and I think from my perspective, I mean it is about us building tools for the customer so that we can simplify the whole process for them. So that these multi-cloud can be treated as another site. I mean, whether you are deploying it on-prem, whether you're deploying in AWS or Azure, these are different sites to you, and you don't have, as a user, have to worry about the nuances of AWS, versus Azure, versus IBM, versus on-prem. You should be able to say that this is my intent, deploy it in AWS, deploy it in on-prem, and be able to move the workloads accordingly. So if I extract what you guys just said is, if the hybrid and cloud equation operationally solves itself, technically with software and automation, all that stuff, the business issues, the app development, basically the app, drive everything. Absolutely. That's a good summary. That's the nirvana, I mean, how far, are we going to hear some of that on the show this week? Absolutely. I think you're going to hear some of these pieces, actually. How are we tying together, and this is like, how are we tying together business intelligence, with infrastructure intelligence? I think you're going to hear some of it. And the good trend for the data center business is that the edge can look like a data center too. The data center is everywhere the data is. That is our mantra, and so that means we're everywhere. Okay. Thanks for coming on theCUBE, really appreciate your insights. Great to have you on. Thanks for joining us. Appreciate it. Thank you very much. Appreciate it, thank you. I'm Trevor Stu Miniman. This is theCUBE kicking off day one. Cisco Live 2020 in Barcelona, Spain. Thanks for watching.