 From around the globe, it's theCUBE with digital coverage of VMworld 2020, brought to you by VMware and its ecosystem partners. Hello everyone, welcome back to the VMworld 2020 virtual coverage with theCUBE virtual. I'm John Furrier, Dave Vellante, your hosts, our 11th year covering VMworld. Got a great guest, Greg Lavender, SBP and the CTO of VMware. Welcome to theCUBE virtual for VMworld 2020 virtual. Greg, thanks for coming on. You have a pleasure to be here, thank you. Really, one of the things Dave and I were commenting with Pat and just in general, it's our 11th year covering VMworld. Absolutely different, it's not face to face. But there's always been a technical conference, always a lot of technical innovation. Project Monterey's out there, it's pretty nerdy, but it's, I call it the catnip of the future, right? People getting excited by it, right? So there's really a lot of awareness to it because it kind of, it smells like a systems overhaul, it smells like an operating system, feels like a lot of moving parts that are quite frankly what distributed computing, geeks and software geeks love to hear about. And to end, distributed software intelligence with new kinds of hardware innovations from NVIDIA and whatnot. Where's that innovation coming from? Can you share your thoughts on this direction? Yeah, I think first I should say, this isn't like something that just, we decided to do six months ago, actually in the office of the CTO four years ago, we actually had a project, future looking project to get our core hypervisor technology running on ARM processors. And that incubated in the office of the CTO for three years, and then last December, I moved the engineering team that had done that research and advanced development work in the office of the CTO over to our cloud platforms business unit, and smart NICs kind of could converge with that. And so we were already well along the innovation path there and it's really now about, building the partnerships we have with the smart NIC vendors and driving this technology out to the benefit of our customers who want to leverage it. You're giving- Greg, I wonder if you could clarify something for me on that. So Pat talked about Monterey and a complete re-architecting of the IO stack. And he talked about effecting NVIDIA and Intel, Melanox, and Sandos part of that. When he talks about the IO stack, specifically, what are we talking about there? So, you know, any computing server in the data center, you know, in a COLA facility or even in the cloud, you know, a large portion of the, you know, CPU resources and even some memory resources can get consumed by just processing, you know, the high volumes of IO that's going out to, you know, storage devices, you know, communicating between the different parts of multi-tiered applications. And so there's an overhead that gets consumed in the core server, CPU, even if it's multi-core, multi-socket. And so by offloading a lot of that IO work onto the ARM core and taking advantage of the hardware offloads that are in those smartNICs, you can offload that processing and free up even as much as 30% of the CPU of a server, multi-socket, multi-core server and give that back to the application so that the application gets the benefit of that extra compute and memory resources. So what about a single sort of low cost flash tier to avoid the complexities of tiering? Is that part of the equation? Well, you know, you can, you know, much storage now is network attached. And so you could, if it's all flash storage, you know, using something like NVMe fabric over ethernet, you can essentially, you know, build large-scale storage networks more efficiently, you know, more cheaply and take advantage of that offload processing to begin to reduce the IO latency that's required to access that network attached storage. And not just storage, but, you know, other devices that you can use that are network attached. So disaggregated architectures is firm. Is that a yes or is that a stay tuned? Yeah, yeah, yeah, yes. I mean, the storage, you know, more efficient use of different classes of storage and storage tiering is definitely a prime use case there. Yeah, great, thank you. Thanks for that, John. How could people think about the Edge now? Because one of the things that's in this end-to-end is the Edge, Pat brought it up. Multi-cloud and Edge are two areas that are extending off cloud and hybrid. What should people think about the innovation equation around those things? Is it these offload techniques? What specifically in the systems architecture do you guys see as the key keys there? So, you know, Edge is a very diversified heterogeneous place in the architectures of a multi-cloud services. So one thing we do know is, you know, workload, I like to say workload follows data. And a lot of the data will be analyzed and processed at the Edge. So the more that you can accelerate that data processing at the Edge and apply some machine learning inferencing at the Edge, we're almost certainly gonna have Kubernetes everywhere, including the Edge. So I think you're seeing a convergence of the hardware architecture, the Kubernetes control plane and services and machine learning workloads, you know, traveling to the Edge where the date is going to be processed and actions can be taken autonomously at the Edge. So I think we're at this convergence point in the industry where all that comes together. How important is the... Do you see that? Oh, God, John. How important is the intelligence piece? Because, again, the data's at the Edge, how do you guys see the data architecture being built out there? Well, again, it's depending on, you know, the thick Edge or the thin Edge, you know, it's going to have different types of data. And again, a lot of the inferencing that could happen at the Edge is going to, I think, more, you know, again, to take action at the Edge as opposed to calling home to a cloud, you know, to decide what to do. So depending on, you know, the computational power and the problem with its video processing or monitoring, you know, sensors or an oil well, the kind of inferencing that will happen at the Edge will be dependent on that data type and what kind of decisions you want to make. So I think data will be moving, you know, from the Edge to the cloud for historical analytics and maybe, you know, additional training mechanisms. But, you know, the 5G is going to play heavily into this as well, right, for the network connectivity. So we're at this unique point as often occurs in the industry every few years. Of all these technology innovations converging to open up an entirely new platform in a new way of computing that happens at the Edge, not just in your data center or at the cloud. So Greg, you did a fairly major stint at a large bank. What would something you mentioned, you know, like an oil rig, but what would something like these changes mean for an industry like banking or financial, will it have an impact there? Put on your customer hat for a minute and take us through that. I mean, you know, ATM machines, you know, branches, kiosks, you know, there's all, banks always been a very distributed computing platform. Right. And so, you know, people want to deliver more user experience services, more video services, you know, all these things at the Edge to interact positively with the customer without using people in the loop. And so the banking industry has already gone through the SD-WAN and I-WAN transformation to deliver the bandwidth more capably to the Edge. And I just think that they'll just now be able to deliver more Edge services that happen, can happen more autonomously at the Edge, as opposed to having the hairpin, the home run, everything back to the data center. Awesome. We have Pat talks about the modern platform, the modern companies. Greg, I want to ask you, because we're seeing with COVID, there's two use cases, you know, the people who don't have a tailwind, companies that are, you know, not doing well because there's no business that we have, they're re-modernizing their business while they have some downtime. Other ones have a tailwind. They have a modern app that, that takes advantage of this COVID situation. So that brings up this idea of, what does a modern app look like? Because now if you're talking about a distributed architecture, some of the things you're mentioning around, inference, data, Edge, people are starting to think about these modern apps and they are changing the game for the business. Now you have vertical industries, you mentioned oil and gas, you've got financial services. It used to be, you had an industry solution, it worked like that and was siloed. Now you have a little bit of a different architecture. If we believe that we're looking up, not down, does it matter by industry? How should people think about a modern application? How do they move faster? Can you share your insights into, into some of this conceptual, what is a modern approach? And does it, does it matter by vertical or industry? Yes, I mean, certainly over the course of my career, I mean, there's just a massive diversity of applications and of course, you know, the explosion of mobile and Edge computing is, you know, just another sort of set of use cases that will put demands on the infrastructure and the architecture and the networking. So a modern app, I mean, we historically built sort of these monolithic apps where we sort of built these sort of three tier apps with, you know, sort of the client side, the middleware side, the database, the backend as the system of record. I mean, this is even being more disaggregated in terms of, you know, the consumer edge is both not just web tier, but mobile tier and, you know, we'll see what emerges out of that. The one thing for sure that is that they're becoming less monolithic and more a conglomeration of SaaS and other services that are being brought together, whether it's from the cloud services or whether it's ISVs delivering, you know, bring your own software and they're becoming more distributed because people need operated higher degrees of scale. There's a limit to vertical scaling. So you have to go to horizontal scaling, which is what the cloud is really good at. So I think all of these things are driving a whole new set of technologies like next generation API gateways, message buses, service mesh, we're announcing Tansu service mesh at VMworld, you know, it's just allowing it's allowing that application to be disaggregated and then integrated with other apps or SaaS services that allow you to get faster time to market. So speed of delivery is everything. So modern CI CD, modern software technology and the ability to deploy and run that workload anywhere at the edge of the core in the data center in the cloud. So when you do your rearchitecture like this, Greg, I mean, you've seen over the course of history in our industry, you've seen so many companies have hit a wall and in VMware as it's just amazing engineering culture, how are you able to, you know, change the engine mid-flight here and avoid like serious technical debt. And I mean, it took, you said, it started four years ago, but can you give us a peek inside, you know, that sort of transformation and how you're pulling that off? Well, I mean, we're delivering the platform and you know, Spring Boot is a key, you know, technology that's used widely across the industry already, which is what we've got as part of our pivotal acquisition. And so what we're just trying to do is just keep delivering the technology and the platform that allows people to go faster with quality, security and safety and resiliency. That's what we do really well at VMware. And so I think you're seeing more people building these apps cloud native as opposed to, you know, taking an existing legacy app and trying to refactor it. They might do what I call, I think somebody's called two-speed architectures, take the user front end, the consumer front end and put that cloud native in the cloud, but the backend system of record still runs in the private cloud in a highly resilient, you know, backed up disaster-recovered way. So you're having, I think, brand new cloud native apps we're seeing and then you're seeing people very carefully because there's a cost to it of looking at how do I basically modernize the front end, but maintain the reliability, the scalability, the security and the reliability of that sort of system of record back end. And so either way, it's winning for the companies because they can do faster delivery to their businesses and their clients and their partners, but you have to have resiliency and reliability that we're known for, running those mission-critical workloads. Right, and so the scenario is that backend stays on-prem and on the last earnings call, I think Pat said or somebody said that I think, as they said, on-prem or maybe they meant hybrid, 30 to 40% cheaper than doing it in the cloud. I presume they were talking about those kind of backend systems that you don't want to migrate. Can you add some color to that again from your customer perspective, the economics? Yeah, you know, somebody asked me one time, what's really a cloud Greg? And I said, it's automation, automation, automation. You can take your current environments and highly automate the release, lifecycle management, develop more agile software delivery methods. And so therefore you can get sort of cloud benefits from your existing applications by just highly optimizing them and on the cost of goods and services. And then again, the hybrid cloud model just gives customers more choice, which is, okay, I want to reduce the number of data centers I have, but I need to maintain reliability, scalability, et cetera, take advantage of the hybrid cloud that we offer, but still run things cloud-native. So I think you're seeing this true multi-cloud technology and paradigm grow out as people have these choices. And then the question is, okay, if you have those choices, how do you maintain security? How do you maintain reliability? How do you maintain uptime? Yet be able to move quickly. And so I think there's different speeds at which those platforms will evolve. And our goal is to give you the ability to basically make those choices and optimize for economics as well as technical capability. Great, I want to ask you a question with COVID. We're seeing, and we've been reporting, the CUBE virtual evolved because we used to be at events, but we're not there anymore. But as everyone has realized with COVID, it's exposed some projects that you might not want to double down on or highlighted some gaps in architecture. I mean, certainly who would have forecast the disruption of 100% work from home, VPN provisioning to access and access management, security. And it really has exposed what kind of who's where in the journey, and digital transformation. So I got to ask you, what's the most important story or thing to pay attention to as the smart money and smart customers go, hey, you know what? I'm going to double down on that. I'm going to kill that project or sunset that or I'm not going to refactor that. I'm going to containerize it. And there's probably, there's a lot of that going on in our conversations with customers. They're like, it's pretty obvious. It's critical path. It's like, we stay in business, we build a modern app, but I'm doubling down. I'm transitioning. It's a whole nother ball game. What is the most important thing that you see that people should pay attention to around maintaining an innovation and coming out on the other side? Yeah, well, I think, I think it just generally goes to the whole thesis of software defined. I mean, you know, the idea of taking on appliance, physical, you have to order the hardware, get it on your loading dock, install it in your data center, go configure it, map it into the rest of your environment, or you can just spin up new software instances of load balancers, firewalls, et cetera. So I think, what's really helped in the COVID era is the maturity of software defined, everything, compute storage, networking, WAN really allowed customers and many of our customers to rapidly make that pivot. And so, what is the workspace, the remote workspace you got to secure it, that's a key part of it. And you've got to give it, you got to have the scalability back in your data centers, or if you don't have it be able to run those virtual desktops in the cloud. And I think, so this ability again to take your current environment and more importantly, your operating model, which the technology could be agile and fast, but if your operating model is not agile, you can't execute it well. One of the best comments I heard from a customer's CIO was, for six months we debated the virtual networking architecture and how to deploy the virtual network. And when COVID hit, we made the decision and did it all in one week. So the question the CIO asked now is like, well, why do we have to operate in that six month model going forward? Let's operate in the one week model going forward. Why did we do that before? I think that's the big inflection point is the operating model has to be agile. We got all kinds of agile technology and choices I mentioned. It's like, how do you make your organization agile to take advantage of those technological offerings? And that's really what I've enjoyed the last six months helping our customers achieve. I think that's a key point worth calling out and doubling down on Dave, because whether you talk about our cube, cube virtual, our operating model has changed and we're doing new things, but it's not bad, it's actually beneficial. We can talk to more people. This idea of virtualization, I mean pun intended, virtualizing workforces face to face interactions are now remote. This is a software defined operating business. This is the real innovation. I think this is the exposure. As companies wake up and go, why didn't we do that before? Reminds me of the old mainframe days, Dave. Why don't we have that mainframe? Because they're still clutching and grabbing onto it. They got a transition. So this is the new reality. We were joking earlier that if they ain't broke, don't fix it. And all of a sudden COVID broke everything. And so virtualization becomes a fundamental component of how you respond. And I wonder if Greg, you could talk about the security piece, how that fits in. Everybody, the bromide of course is, security can't be a bolt on. It's got to be designed in from the start. Pat Gelsinger said years ago in the cube, security is a do-over. You guys have purchased many different security components. You've built in security. So how should we think about, and how are you thinking about designing in security across that entire stack without really bolting in pieces, whether it's carbon black or other acquisitions that you've made? Yeah, I mean, I think that's the key inflection point we're in as an industry. I mean, getting back to my part of banking experience, I was responsible for cybersecurity engineering, the platforms that we engineered and deployed across the bank globally. And the challenge, let's say I had 150 plus security products and you go to bed at night wondering, which one did I forget to deploy or what did I get? But it's for gaps. You think you're safe by the sheer number, but when you really bowl down to it, it's like, you know, cause you have to sort of like bolt all this stuff together to create a secure environment, you know, on a global level. And so really our philosophy at VMware is, okay, well, let's kind of break that model. Let's, we call it intrinsic security, which is just, you know, we have the hypervisor. If you're running the hypervisor, it's running on most of the servers in your data center. If you have our network virtualization, we see all the traffic going between all those hypervisors and out to the cloud as well, and then we have the hyper cloud or public cloud with our NSX technology. And then, you know, then you sort of bring in to that the load balancers and the software defined firewalls and pretty soon you have, you realize, okay, look, we have most of the estate, therefore we can see everything and bring some intelligent machine learning to that and get proactive as opposed to reactive. Because our whole model now is we have this technology and some alert pops and we get reactive about proactively telling me that something nasty is going on. I need to ask you a question. And maybe remediate it, sorry, John, and maybe remediate it, you know, at some point anyway, bring in some machine intelligence to instead of, like you said, get an alert, actually tells me what happened and how it was fixed, you know, or at least recommending what I should do. Right? Yeah, part of the problem in the historic architectures is it was all these whole silos, you know, every business unit had its own sort of technology and as you make things virtualized, you sort of do the virtual networking, the virtual storage, the virtual compute, all the software, you know, all of a sudden you have a different platform, you have lots of standardization, therefore you're operating model simplifies, right? And then it's about just collecting all the data and then making sense of the data so that you're not overwhelming the human's capacity to respond to it. And so I think that's really the fundamental thing we're all trying to get to, but the surface area is enlarged outside the data centers. We've discussed out to the edge, whatever the edge is, you know, into the cloud, hybrid or public. So now you've got this big surface area where you've got to have all that telemetry and all that visibility. Again, back to getting proactive, so you got to do it in-band as opposed to out-of-band. Great, I want to ask you a question on cybersecurity. We have an event on October 4th, a virtual event. The Cube is hosting with Cal Poly around the space and cybersecurity symposium as intersection of space and cyber. I noticed VMware recently announced last month that the United States Space Force has committed to the Tanzu platform for continuous DevOps operation for agility. I interviewed Lieutenant General John Thompson, Space Force, and we talked about that. He said, quote, it's hard to do break-fix in space. Illustrating really, you just can't send someone to swap out something in space, not yet, at least. So they're looking at software defined as a key operating reality. Okay, so again, talk about the edge, the space is as edge as you're going to get. It needs to be completely managed. Talk about payloads and data. This is kind of an interesting data point because you have security issues because space is going to be contested and congested as an edge device. So it's obviously the government's interested in that. But fundamentally, it's a DevOps problem that you guys are involved in. This is the reality. This kind of connects this reality idea of operating models based in reality have to be software. What's your take on it? Yeah, I mean, I think the term we use now is DevSecOps because you can't just do DevOps. You have to have the security component in there. So yeah, the interesting, there's a lot of interesting things happening just in fundamental networking, right? I mean, the Starlink satellites that Tesla has launched or Elon Musk has launched and bringing sort of higher bandwidth, lower latency to those, we'll call it near space. And then again, it just opens up all new opportunities for what we can do. And so yeah, I think the whole software development ecosystem, again, back to this idea, I think of three things. You gotta have speed, you gotta have scale, and you gotta have security. And so that's really the merging platform, whether it's terrestrial or in near space, that's giving us the opportunity to do new architectures, create service meshes of services, some terrestrial, some far remote. And as you bring these new application architectures and system platform architectures together with all the underlying hardware and networking innovations that are occurring, you mentioned flash, but even we're getting into P and M persistent memory, right? So this is so much happening that is converging. And what's exciting to me about being at VMware as the CTO, we partner with all the hardware vendors, we partner with all the system providers like NVIDIA and others, the smart NIC vendors. And then we get to come up with the software architectures that sort of bring that together holistically and give people a platform where you can run your workloads to get work done wherever you need to land those workloads. And that's really the excitement about the software. You're a kid in the candy store and yet you got hard problems to work on to solve. I mean, this really brings the whole project monorateful circle. Because we think about space and networks and all these things you're talking about, you need to have smart everything. I mean, isn't that software, it's a complete tie-in to the monorail. Yeah, exactly, you're right. It's not just connecting everything and pushing data around. It's then having the intelligence to do it efficiently, economically and securely. And that's, so I see the, I don't want to over hype machine learning. I tend not to use the term AI, but use the machine learning technologies, properly trained with the proper data sets, and then the proper algorithms, that you can then deploy at the edge, the small edge, thick edge, in the data center, at the cloud. It's really then you give the visibility so that we get to that proactive world I was talking about. Yeah, great stuff. Greg, great insight and great conversation. Looking forward to talking more tech with you. Obviously you are in the right spot of VMworld in the center of all the action across the board. Final point, if you can just close it out for us, what is the most important story at VMworld 2020 this year? Well, I think, I like to say that I have the best job, I think that I've had in my career, and I've had some great ones, is we get to be disruptive innovators, and we have a culture of perpetual innovation. And really VMworld for us, all the employees and all the people that work together to put it together, is we get to showcase some of that. Obviously I have more up our sleeves for the future, but VMworld is our coming out show of the latest set of innovations and technology. So there's going to be so much. I have a vision and innovation keynote kickoff where I do some lightning demos, and actually I talk about work we're doing in sustainability, and we're putting a microgrid on our campus in Palo Alto in partnership with the city of Palo Alto, so that when the wildfires come through or there's power outages, we're an oasis of power generating capacity with our solar and our batteries. And so the city of Palo Alto could take their emergency command vehicles and plug into our batteries when the power is out in Palo Alto and operate city services and city emergency services. So we're not just innovating in core tech, we're innovating to become a more sustainable company and provide sustainable carbon neutral technology for our customers to adopt. And I think that's an area we want to talk about, maybe we can talk about it next time, but I think our innovations are going to basically help change the world with regard to climate as well. Let's definitely do that. Let's follow up for another in-depth conversation on the societal impact. Of course, VMware is VMworlds 2020. It's virtual, there's a ton of sessions. There's a cloud city portion, check out the 60 solution demos. Of course, they asked the expert, Greg, you're in there with Joe Beda, Raghu, all the experts, engage and check it out. Thank you so much for the insight here on theCUBE, virtual. Thanks for coming on. Appreciate the opportunity, great conversation and good questions. Great stuff, thank you very much. Thanks for it. Innovation at VMware, it's the heart of their missions always has been, but they're doing well on the business side, Dave. Okay, theCUBE coverage there, not there in person, virtual. I'm John Furrier, Dave Vellante. Thanks for watching.