 Welcome back to San Francisco. theCUBE is live, our day two coverage of VMware Explorer 2022 continues. Lisa Martin with Dave Nicholson. Dave and I are pleased to welcome Jason Collier, principal member of technical staff at AMD to the program. Jason's great to have you. Thank you, it's great to be here. So what's going on at AMD? I hear you have some juicy stuff to talk about. Oh, we've got a ton of juicy stuff to talk about. The, you know, clearly the project monitoring announcement was big for us, right? So we've got that to talk about. Another thing that I really wanted to talk about was a tool that we created, and we call it, it's the VMware architecture migration tool. Call it VAMP for short. It's a tool that we created and we worked together with VMware and some of their professional services crew to actually develop this tool. And it is also an open source based tool. And really the primary purpose is to easily enable you to move from one CPU architecture to another CPU architecture and do that in a cold migration fashion. So you're probably not talking about CPUs from Tandy Radio Shack systems. Likely this would be what we might refer to as other x86 systems. Other x86 systems is a good way to refer to it. So it's interesting timing for the development and the release of a tool like this because in this sort of x86 universe, there are players who have been kind of delayed in terms of delivering their next gen stuff. My understanding is AMD has been public with the idea that they're on track for by the end of the year, Genoa. Right, next gen architecture. So can you imagine a situation where someone has an existing set of infrastructure and they're like, hey, you know what? I want to get on board the AMD train. Is this something they can use from the VMWare environment? Absolutely, and when you think about it. Tell us exactly what that would look like. Walk us through, you know, 100 servers, VMWare, 1000 VMs, just make the math easy. What do you do? How's it work? So one, there are several things that the tool can do. We actually went through, the design process was quite extensive on this. And we went through all of the planning phases that you need to go through to do these VM migrations. Now this has to be a cold migration. It's not a live migration. You can't do that between the CPU architectures. But what we do is you create a list of all of the virtual machines that you want to migrate, right? So we take the CSV file, we import the CSV file, and we ask for things like, okay, what's the name? Where do you want to migrate it to? So what, from one cluster to another, do you want to migrate it to? What are the networks that you want to move it to? And then the storage platform. So we can move storage. It can either be shared storage or we can move, say, from like vSAN to vSAN, however you want to set it up, right? So it will do those storage migrations as well. And then what happens is it's actually gonna go through, it's going to shut down the VM. It's gonna take a snapshot. It is going to then basically move the compute and or storage resources over. And once it does that, it's gonna power them back up and it's going to check. It's gonna, we've got some validation tools where it's going to make sure VM tools comes back up where everything is copaesthetic, right? It didn't blue screen or anything like that. And once it comes back up, then everything's good, it moves on to the next one. Now a couple of things that we've got feature-wise we built into it, you can parallelize these tasks. So you can say, how many of these machines do you want to do at any given time? So it could be, say 10 machines, 50 machines, 100 machines at a time that you want to go through and do this move. Now, if it did blue screen, it will actually roll it back to that snapshot on the origin cluster. So that there is some protection on that. Couple other things that are actually in there are things like audit tracking. So we do full audit logging on this stuff. We take a snapshot, there's basically kind of an audit trail of what happens. There's also full logging, sys logging, and then also we'll do like email reporting. So you can say, run this and then shoot me a report when this is over. Now one other cool thing is you can also actually define a change window. So I don't want to do this in the middle of the afternoon on a Tuesday, right? So I want to do this kind of later at night on over the weekend. You can actually just queue this up, set it, schedule it. It'll run, you can also define how long you want that change window to be. And what it'll do, it'll do as many as it can. Then it'll effectively stop, finish up, clean up the tasks, and then send you a report on what all was successfully moved. Okay, I want to go down the rabbit hole a little bit on this. Because I think it's important. And if I say something incorrect, you correct me. In terms of my technical understanding. So you've got a VM, essentially a virtual machine typically will consist of an entire operating system within that virtual machine. So there's a construct that containerizes, if you will, the operating system. Where, what is the difference? Where is the difference in the instruction set? Where does it lie? Is it in the OS's interaction with the CPU? Or is it between the construct that is the sort of wrapper around the VM that is the difference? Yeah, it's really primarily the OS, right? And this is, you know, we've not really had too many issues doing this. And most of the time what's going to happen that OS is going to boot up. It's going to recognize the architecture that it's on. It's going to see the underlying architecture and boot up. All the major operating systems that we test, you know, work fine. I mean, you know, typically they're going to work on, you know, all the x86 platforms. But there might be instruction sets that are kind of enabled in one architecture that may not be in another architecture. But you're looking for that during this process. Well, usually the OS itself is going to kind of detect that. So if it pops up, the one thing that, you know, that is kind of a caution that you need to look for, if you've got an application that's explicitly using an instruction set that's on one CPU vendor and not the other CPU vendor, that's the one thing where you're probably going to see some application differences. That said, it'll probably be compatible, but you may not get that instruction set advantage in it. Okay, but this tool remediates against that. Yeah. So, and what we do, we're actually using VM tools itself to go through and validate, you know, a lot of those components. So we'll look and make sure VM tools is enabled in the first place on the source system. And then when it gets to the destination system, we also look at VM tools to see what is, what is and what is not enabled. Okay, I'm going to put you on the spot here. What's the zinger? Where doesn't it work? You already said cold, we understand, you can schedule for cold migrations. So that's not a zinger. What's the zinger? Where doesn't it work? It doesn't work, like live migrations just don't work. No live? Okay, okay, fine. What about something else? What's the, oh, you've got that version. You've got that version of X86 architecture. Yeah. It could work. Anything? The majority of those cases work where it would fail, where it's going to kick back and say, hey, you know, it's VM tools is not installed. So where you would see this is like, if you're running a virtual, like a virtual appliance from some vendor. Okay. You know, you like insert vendor here that say got a firewall or got, you know, something like that. And they don't have VM tools enabled. It's going to fail it out of the gate and say, hey, VM tools is not on this. You might want to manually do it, right? But you can figure out how to fix that. You can figure out how to do that. We can also, and there is a flag in there. So in kind of the options that you give it, you say that ignore VM tools, don't care, move it anyway. So if you've got, you know, less, you know, like some VMs that are in there, but they're, you know, not a priority VM, then it's going to migrate just fine. Can you talk elaborate a little bit on the joint development work that AMD and VMware are doing together and the value in it for customers? Yeah. I mean, so it's one of those things we worked with VMware to basically produce this open source tool. So we did a lot of the core component and design and we actually engaged VMware professional services, you know, a big shout out to Austin Browder. He helped us a ton in this project specifically and we basically worked, we created this kind of code design, what it was going to look like and then kind of jointly worked together on the coding, you know, of pulling this thing together. And then, you know, after that, so this is actually posted up on, you know, VMware's public repos now in GitHub. So you can go to GitHub, you can go to the VMware samples code and you can download this thing that we've created. And, you know, it's really built to help ease migrations, you know, from one architecture to another. So if you're looking for a big data center move and you got a bunch of VMs to move, I mean, even if it's, you know, like same architecture to same architecture, it's definitely going to ease that, you know, kind of the pain of going through and doing a migration of, you know, it's one thing when you're doing 10 machines but when you're doing 10,000 virtual machines, that's a different story. It gets to be quite operationally inefficient. I lose track after three. Yeah. So I'm good for three, not for. I was going to ask you what your target market segment is here, expand on that a little bit and talk to me about who you're working with and those organizations. Right, so really this is targeted toward organizations that have, you know, large deployments and enterprise, but also I think this is a big play with channel partners as well. So folks out there in the channel that are doing these migrations and they do a lot of these for when you're thinking about the small and mid-sized organizations, it's a great fit for that, especially if they're, you know, kind of doing that, that upgrade, the lift and shift upgrade from like, here's you've been five to seven years on an architecture and you want to move to a new architecture, this is really going to help and the, this is not a point and click GUI kind of thing, right? It's a, you know, command line driven it's using PowerShell or using PowerCLI to do a majority of this work and for channel partners, this is an excellent opportunity to put the value in the value add of a var, right? And there's a lot of opportunity for I think channel partners to really go and take this and once again, be an open source, we expect this to be extensible, we want the community to contribute and, you know, kind of put back into this to basically help grow it and make it a more useful tool for doing, you know, these cold migrations, you know, between CPU architectures. Have you seen any in the last couple of years of dynamics, obviously, across the world? Any industries in particular that are really leading edge for what you guys are doing? Mm, that's a, yeah, that's really interesting. I mean, we've seen it, it's honestly been a very horizontal problem, pretty much across all vertical markets. I mean, we've seen it, you know, in, you know, financial services, we've seen it in, you know, my honestly, you know, pretty much across the board. I mean, I think manufacturing, manufacturing, financial services, healthcare, we have seen, you know, kind of a strong interest in that. And then also we've actually taken this and presented this to some of our channel partners as well. And there was a lot of, there's been a lot of interest in it. I think we presented to about 30 different channel partners a couple of weeks back about this and I got contact from 30 different channel partners that said they're interested in basically helping us work on it. So. You know, tagging on to Lisa's question, do you have visibility into kind of the AMD thought process around the timing of your next gen release? Right. Versus others that are competitors in the marketplace? Right. How you might leverage that in terms of programs where partners are going out and saying, hey, perfect time, you need a refresh, perfect time to look at AMD. Yeah. If you haven't looked at them recently. Do you have any insight into that and what's going on? I know you're focused on this area. Yeah. But what are your thoughts? I mean. What's the buzz? What's the buzz inside AMD on that? Well, when you look overall, when, I mean, if you look at like the Gartner hype cycle when VMware was being broadly adopted, right? When VMware was being broadly adopted, I'm going to be blunt and I'm going to be honest right here. AMD didn't have a horse in the race, right? And the majority of those VMware deployments we see are not running on AMD. Now that said, there's an extreme interest in the fact that we've got these very core dense systems that are now coming up on, down here at that five to seven year refresh window of pulling in new hardware. And we have extremely attractive hardware when it comes to running virtualized workloads. Test cluster that I'm running at home, I've got that five to seven year old gear and I've got some of the, even just the Milan systems that we've got. And I've got three nodes of another architecture going on to AMD. And when I got that system, these three nodes completely maxed to the number of VMs that I can run on them. I'm at a quarter of the capacity of what I'm putting on the new stuff. So what you get is, we've worked the numbers and it's definitely, it's like a 30% decrease in the amount of resources that you need. Which is, that's a compelling number. It's a compelling number. 5%, 10%, nobody's going to do anything for that. You talk 30%. 30%. It's meaningful, it's meaningful. Now you're out of Austin, right? Yes. So first thing I thought of when you talk about running clusters in your home is the cost of electricity, but you're okay. I'm okay. You don't live here. You don't live here, you don't need to worry about it. I'm okay. Do you have a favorite customer example that you think really articulates the value of AMD when you're in customer conversations and they go, why AMD? And you hit back with this. Yeah, actually it's funny because I had a conversation like that last night. You know, random, you know, kind of a random person I met later on in the evening. We were going through this discussion and they had, they were facing exactly this problem. They had that five to seven year infrastructure. It's funny because the guy was like a gamer too. And he's like, man, I've always been a big AMD fan. I love, I love the CPUs. I've, you know, all the way since you know, you know, back in basically the, the Optorons and Aflons, right? He's like, I've always loved the AMD systems. Love the graphics cards and now with the, you know, what we're doing with Ryzen and all that stuff. He's, I've always been a big AMD fan. He's like, and I'm going through doing my infrastructure refresh. And I told him just like, well, you know, hey, pop the talk to your, talk to your var and, and, you know, have him plug some AMD SKUs in there from, from the Dell's HP's and Lenovo's. And then we've got this tool to basically help make that migration, you know, easier on you. And so, you know, once we, we, we had that discussion and it was great. Then he swung by the booth today and I was able to, you know, kind of just go over like, Hey, this is the tool. This is how you use it. Like, here's all the info. Call me if you need any help, right? So yeah, we, when we were talking earlier, we learned that you were at scale. Yes. So what do you, what do you, what do you like it about AMD? How does, how does, how does that relate? The funny thing is this is like actually the first time in my career that I've actually had a job where I didn't work for myself. I've been doing, I've been doing venture-back startups the last 25 years and we've raised like, you know, a couple hundred million dollars worth of, worth of investment over the years. And so one, I figured like, here I am going to AMD, a larger corporation. I'm just like, am I going to be able to make it, you know, a year, right? And I have been here longer than a year and I absolutely love it. The culture at AMD is amazing. We still have that really, I mean, almost it's like that, that underdog mentality within the, within the organization. And the team that I'm working with is a phenomenal team and it's actually, you know, our, our, the kind of, our EVP and our Corp VP were actually my executive sponsors. They were at a prior company. They were, they were one of my executive sponsors when I was at scale. And so, you know, my, my now VP boss calls me up and says, hey, I'm putting a band together. Are you interested? And I was like, I was kind of enjoying a semi-retirement, you know, lifestyle. And then I'm just like, man, because it's you, yes, I am interested. And you know, the group that we're in, the work that we're doing, the way that we're, you know, really focusing on forward-looking things that are affecting the data center, what's going to be the data center like three to five years from now? It's exciting, right? And I am having a blast. I'm having the time of my life. Absolutely love it. Well, that relationship and the trust that you will have with each other, that bleeds into the customer conversations, the partner conversations, the employee conversations. It's all inextricably linked. Yes, it is. And we want to know, you said three to five years out. Like what? Like what? What do you think? Yeah. Just general futurist stuff. Where do you think this is going? Well, that's interesting. So Moon collides with the Earth in 2025. We already know that. So we dialed this back to kind of that, the Pensando acquisition. When you look at the Pensando acquisition, that's, and you look at basically where data centers are today. But then you look at where basically the big hyperscalers are, right? You look at an AWS, you look at their architecture. You specifically wrap Nitro around that. That's a very different architecture than what's being run in the data center. And when you look at what Pensando does, that's a lot of starting to bring what these real clouds out there, what these big hyperscalers are running into the grasp of the data center. And so I think you're going to see a fundamental shift. The next 10 years are going to be exciting because the way you look at a data center now when you think of what CPUs do, what shared storage, what's, you know, how the networking is all set up, it ain't going to look the same. Okay, so the competing vision with that to play devil's advocate would be, DPUs are kind of expensive. Why don't we just use Nix, give him some more bandwidth and use the cheapest stuff? Is that, is it, that's the competing vision? That could be. Or the alternative vision. And I imagine like everything else we've experienced in our careers, they will run in parallel paths fit for function. What do you think? Well, parallel paths always exist, right? You know, otherwise, because you know, how many times you've heard mainframes dead, tapes dead, spinning discs is dead, none of them are dead, right? The reality is you get to a point within an industry where, you know, it basically goes from instead of a growth curve like that, you know, it goes to a growth curve like that, right? It's pretty flat, right? So from a revenue growth perspective, I don't think you're going to see the revenue growth there. I think you're going to see the revenue growth in DPUs. And when you actually take, they may be expensive, you know, now, but you look at what Monterey's doing and you look at the way that, you know, kind of those DPUs are getting integrated in at the OEM level, it's going to be a part of it. You're going to, you know, you're going to order your, you know, like VX Rail, you know, and VSAN style boxes, they're going to come with them, right? It's going to be an integrated component because when you start to offload things off the CPU, you've driven your overall utilization up, right? When you don't have to process like NSX on basically the X86, you've just freed up cores and a considerable amount of them, right? And you've also moved that to where there's a more intelligent place for that pack to be processed, right? Out here on this edge, because you know what? That might not need to go into the host bus at all. So you have like just alleviated any transfers like over a PCI bus, over the PCI lanes and DRAM, you know, all of these components when you're like, but that was all to come with like, oh, that bit needs to be on this other machine, right? So now it's coming in and it's making that decision there. And then you take and integrate that into like things with like the Aruba smart switch that's running the Pensando technology. So now you got top of rack that is already making those intelligent routing decisions on where packets really need to go. Jason, thank you so much for joining us. I know you guys could keep talking. You're going to say you're going to have to come back. We've just started to peel the layers of the onion, but we really appreciate you coming by the show talking about what AMD and VMware are doing, what you're enabling customers to achieve. Sounds like there's a lot of tailwind behind you. That's awesome. Yeah. Great stuff. Thank you. It's a great time to be at AMD. I can tell you that. Oh, it's good to hear. We like it. Well, thank you again for joining us. We appreciate it. For our guests and Dave Nicholson, I'm Lisa Martin. You're watching theCUBE live from San Francisco VMware Explorer 2022. We'll be back with our next guest in just a minute.