Virtualization Is (Still) King

  |  Compiler Team   Virtualización

Compiler • • Virtualization Is (Still) King | Compiler

Virtualization Is (Still) King | Compiler

About the episode

Virtual machines are not just the power behind legacy infrastructure. They are becoming a vital pillar for GPUs, which are integral to large language models.

Maria Bracho, CTO for industries at Red Hat, explains how virtualization, far from being replaced, has adapted to leverage the robust and powerful computing behind artificial intelligence.

Compiler team Red Hat original show

Suscribir

Subscribe here:

Listen on Apple Podcasts Listen on Spotify Subscribe via RSS Feed

Transcripción

If you're doing like a whole OS image, that's gonna be a VM. If you're doing like little piddly things or something else, that's gonna be containers. Can I say piddly? I don't know. You can, you can. And everybody that I know that has been containerizing their software the past decade is laughing so hard. They're—hopefully they're listening. Is Compiler, an original podcast from Red Hat. I'm Emily Bock. And I'm Jennifer Scalf. On this show, we go beyond the buzzwords and jargon, and simplify tech topics. This season, we're covering the fundamentals of IT infrastructure. On today's episode, we're discussing virtualization Okay, so imagine you bought an entire house, and not just a house, but a mansion. Yeah, okay. But when you finally moved in, you only lived in one corner of one room. I'm sure there are very wealthy people out there who have done that. Where is this going? Yeah, no, and I'm sure there are some that do that, but I think to the rest of us, that idea sounds really silly, or at the very least, not very efficient. But I think that's also what enterprise computing looked like before virtualization. Oh, okay. I get it. And you know, because the infrastructure we have now depends so heavily on cloud computing, it's really easy to take the cloud for granted. Absolutely. And before the cloud existed, though, engineers had to solve a physical space and hardware problem, and that's where virtualization came in. So here to explain how is Maria Bracho. She's the chief technology officer for industries at Red Hat. Before VMs, if you were a company, you had to run one application on one physical server. And that works if that application is sort of up all the time, running all the time, and you're getting the most utilization of your physical server. You usually don't need that hardware working for one application all the time. So a lot of those servers, uh, which are expensive, ended up being underutilized, running at, you know, 10% capacity, 20% capacity. Yeah, that's the nail on the head. But for posterity, we'll do it ourselves as well. So Jennifer, what is a virtual machine? Oh, I, I could listen to Maria talk about this all day, 'cause I've been on so many customer situations and calls and, uh, intense situations with her specifically, and I just love how she just broke it all down. Because I would've said, "Oh, virtualization, that's where you're virtualizing the hardware." And the people would look at me and say, "Well, what does that mean?" Because they have just a cell phone in front of them now. Like, "You're virtualizing my cell phone? What does that mean?" Okay, so it's a fake cell phone. Okay. No, no, no. There's way more to it. So for folks who have never been in a server room, because I really don't think m- many people go into server rooms anymore. Mm-hmm. But if you think back to maybe a movie or a TV show where somebody had a whole stack of blinking lights and machines and machine racks, those are the physical machines that we're talking about. We're not talking about virtualizing necessarily a cell phone or a laptop- Mm-hmm ... or, you know, a desktop. Um, these massive data centers full of servers, each one of those, yeah, they used to run, uh, an entire operating system, an entire, um, eh, whatever application folks were looking at. It was very connected, very easily to hold in your hand, ironically. And so once somebody realized, "Oh, well, that doesn't, uh, necessarily scale-" Mm-hmm 'cause we've either run out of room or it's very expensive to just keep buying more RAM or, you know, more racks full of servers. Well, what can we do to better distribute applications across these?" So the first step in that, before folks jumped too far into the future, the first step was taking your bare metal and virtualizing the actual operating systems that you were using so you could run those operating systems across many, many different physical pieces of hardware, or- Actually out into geographically separated locations. Mm-hmm. So you get that level of distributed, high available operating systems, and then your application's on top of those operating systems. So yes, I would've taken a long time to say what Maria said. I didn't mean to laugh at that. That's just kind of our thing. No, right? Like, that's the thing. I'm like, "It's so much more complicated, this topic." But she did such a great job saying, "Yep, physical server, one application, and now you don't, you're not constrained. Um, we've developed all kinds of software that lets you pretend like you have a lot more pieces of hardware than you actually do." Yeah. So, like, a lot of the cloud stuff is kinda like renting space on someone else's server so you don't need the whole thing. Is that kind of the, the concept? Exact- oh, see, and you said it even much more simply. This is what happens when you work in this for so long. You're like, "No, I can't. I'm sorry, I can't make this simple. We're gonna have to go deep." Well, when you know how deep it is, it's hard to hit the shallow spots. So that's what I'm here for. It very much is. So yes, now you know what the cloud is, what the actual cloud is, and 10 years, whenever I started, 12 years ago, 14 years ago, yeah, 14 years ago, when you talked about cloud computing, that is exactly what you meant at the time, was that somebody has virtualized the operating system layer of the stack. Gotcha. That makes sense. And I imagine that comes with some trade-offs too. Like, there's the benefit where you don't need a whole physical server or servers if that's the maximum space you'll ever need, so, uh, better for scaling and more efficient. What would be, like, the, the drawbacks? When would you want a physical server instead? I'm just gonna go right back to security. Mm-hmm. If you need to have it, quote-unquote, air gapped- Mm-hmm ... which is what it's called when, when a server is not online at all, you could have an air gap- gapped cloud. It, it takes some space and some money. It's kind of expensive. It can be to have an entire cloud environment. So if you do not have a lot of money or you wanna have it physically, completely separated from the internet, from the rest of the world, that's a couple of good, good, good times when you would not want to, yeah, virtualize. Super-duper security or, like, super, super internal proprietary kind of information, that kind of a, a thing. But generally speaking, I guess also if you have stuff in the cloud, you are beholden to the cloud. That's a very good point. If there are other users, if you have an outside group that's maintaining it, they can upgrade whenever, um, there might be reasons why you would want to be, have more control over all of that. Yeah. No, I, I think that makes a lot of sense. So how do we get one physical machine to act like 10? You need a traffic cop. Enter the hypervisor. So a hypervisor is a piece of software that sits between the hardware- Which we call the host, and the VMs, which we call the guests. And there can be several VMs on top of that hardware. So that hypervisor allocates physical resources to each VM, making sure that they're completely isolated from one another, you know, so you don't have noisy neighbors, essentially. Noisy neighbors. That's a great analogy. Yeah, it definitely tracks. Hypervisors used to be heavily proprietary and really expensive, and then open source stepped in as, like, a great equalizer of all these things. We have, uh, KVM, kernel-based virtual machines. It effectively turns the operating system itself into a hypervisor. Open source communities around virtualization have ensured that virtualization isn't locked behind a single vendor's paywall. It allows collaborative development to maximize the hypervisor efficiency. So Jennifer, I'm gonna open up a little bit of a can of worms here, 'cause we talked a few episodes ago all about containers, and a lot of the same things came up. So if containers exist, why do we still use VMs? Oh, goodness. So in my line- ... we have to have VMs because, and I'm not putting it on the- Applications or the business owners, but I will for just a moment. Not everything has been containerized. Mm-hmm. So we have software, so I, I'm, I'm still in telecommunications right now, and there's software that goes back a very long time, decades and decades and decades. And honestly, there are reasons why some of them haven't been able to move to containers. Um, I'm not gonna go into the technical details right now, but they have to have an entire operating system, some of these, uh, VNFs, virtualized network functions. Mm-hmm. And that, I mean, that's just simply what it is, right? It's not dictated by us. I work on, a lot on the providing, provider, operator side of the house, and we haven't talked about what operators are yet in the virtual and container world. So we'll get to that too. That's another, another term I'm dropping. We can't leave behind some of the older technology. We have to have an operating system available, but yet we can't be beholden to having one operating system on one piece of hardware, back end server in a server room. It has to be able to scale. You have to be able to have high availability. You have to have all these things, right? So you need to be able to provide a hypervisor that lets these virtual machines run, sometimes in the same environment as containers, sometimes on their own. And I just love, going back to her, um, analogy about the noisy neighbors. There are also times where you want everything an operating system would contain. Oh, now I'm just fuzzing the boundaries of the words right now. But everything an operating system would contain, like unto itself so that it isn't invading its neighbor spaces. Gotcha. So I'm sensing a lot of overlap, but VMs still have some things inherent to them that containers can't as easily fulfill. So if you're doing like a whole OS image- It's gonna be a VM. If you're doing, like, little piddly things or something else, that's gonna be containers. Can I say piddly? I don't know. You can, you can, and everybody that I know that has been containerizing their software the past decade is laughing so hard. They're- hopefully they're listening. But that's what we're trying to get to, right? Mm-hmm. We're trying to get to these applications being torn apart and run in small containers instead of having big beasts of virtual machines with everything that an operating system is. So yeah, we're gonna use that word positively. Um, containers are designed to assume that something is going to fail also. If you have an entire operating system running, the assumption is that it is going to stay up, it is gonna be online. Containers, they go offline, they bleep in and out. They're little fireflies. Oh, we gotta make some more analogies. Fireflies. I just had a mental image of fireflies blipping in and out. When you design your container platform, imagine little fireflies, and they blip off, they bloop off, they blip, blip, blip, and then the other one is on over there. Um, it's much more- It's, like, a little bit more of a permanent concept it sounds like, where, like, a container will blip in and out, but, like, a VM is really pretending to be a real, actual, physical server, 'cause it sort of is somewhere, just not near you. Yes, it is mimicking a physical server, and speaking of fireflies, my cat is now blipping in front, on top- ... of, oh my goodness, on top of my keyboard. We just had a cat interruption. So she's definitely a container. My other cat is a virtual machine. Okay. That explanation makes sense to me. So assuming that our hypothetical person or company or application decides on virtualization, that's the thing they need for what they're doing, what happens next? I don't even know anymore- ... of any other groups that you would u- uh, any other software besides ours that you would use, so I'm really, really biased, but it's deciding then how I know that sounds terrible. It doesn't sound terrible. It's the only one that I know of anymore that's really easy to set up and install and, and get running. But you do need a fair number of, um, CIS admins, DevOps, engineers. It is more complicated to initially set up than bare metal- Hmm ... because you have to decide is everything going to be in the same physical data centers? What is your networking? Yeah. I mentioned VNF's, virtualized network functions. So just like in the, in the past, you would have to find your drivers for your bare metal machine in each one of the, uh, network cards or whatever that you have in there. Now you have to go and you have to set up your, the virtualized version of that. Mm-hmm. And people have to have a pretty good amount of networking background to set that up initially. Once it's set up, you're good. Yeah. But it's that initial set up. Are your cloud nodes all going to be running on the same hardware? You really can connect a lot into this cloud. It doesn't all have to be the same. It doesn't all have to be uniform. That's part of the beauty of it. But, like, sitting down and really taking that inventory, figuring out what will the applications and the operating systems that will be running on top of the hypervisor, what infrastructure does it need now? What are we looking to scale to? Very similar to what you would be doing on a bare metal environment. Mm-hmm. But you, it's different drivers, it's different, uh, and I'm, I'm thinking now of our software has all these amazing names. Neutron, Cinder. I don't know if you remember all these, but there's all these amazing names for the storage, the networking, Ironic. Oh, there's so many great ones. Nova. I can't help myself. Because you get the sense of what each of those does. Mm-hmm. Each of those bare metal pieces of hardware within your environment now are, you have a, a virtualized version of that, and you need people that understand what that is. Yeah. And it can get kinda complicated, I'm not gonna lie to you. Yeah, like maybe even more so in some ways because at least with like a physical server or physical server room, you have kind of the tactile feedback of, okay, this wire goes there, that wire goes there. It's physically here, this is what it looks like. Whereas when you're doing it virtually- so to speak, you have all those same decisions to make, but also you can't see it. Yes, exactly. Oh my gosh. I'm having flashbacks to when I first started learning about this, again, about 14 years ago now. It was really, mm, a little bit earlier I got into virtualization. And as a bare metal person, I went, "I can't see the cable anymore." And it was really painful. So yes, that's exactly right. So it might be a little easier in, like, concept someti- well, at least easier for scaling, 'cause it's much easier to just spin up another VM than to buy and set up a whole other server. Yeah. But that's after it's already, after you've already set up whatever your hypervisor will be. Mm-hmm. It's that I'm thinking about the poor folks that have to learn how to set it up in the first place. Yeah. Oh, yeah. But once they get past that, yeah, you press that button. That's exactly how it's supposed to be. You press that button- Yeah ... you get a new virtual machine, an entire operating system. So you got it, but setting up the equipment to do that is not incredibly easy, so I don't wanna lie to anybody. Yeah. Gotcha. That makes sense. So once it's set up, probably a little easier than maintaining your own servers. Yes. But the setup is just as, if not more difficult, especially if you've never set up a physical one before. Oh, yeah. Oh my gosh, 'cause you have no paradigm. Mm-hmm. Oh, I'm gonna go back to dot com era- jargon, just for a minute. I know we promised no jargon, but if you don't understand what running a bare metal machine looks like, it's very hard, I think, to wrap your head around a virtualized version- Mm-hmm ... right off the bat. But I don't know. Maybe if you don't know any, if you don't know what you don't know, maybe it's easier. Ah, we'll see. Yeah. No, I, I'm a very visual learner, so having the physical thing helps a lot. And trying to just kind of figure out based on text and vibes what I'm doing is a little more difficult. Yes. Oh, and I didn't say it, but I do wanna say it. The original point of virtualization was to optimize CPUs. Mm-hmm. But today's industry, it's all about GPUs. Yeah. So we'll discuss the challenges around virtual machines and their relationship with AI after the break. All right, we're back to virtualization. One of the primary challenges for companies looking at virtual machines is there can be a lot of guesswork around how much is needed Guessing the sizing of a VM, usually, uh, people overshoot. You don't want that to happen. Then the other level would be networking complexity. You need to understand connection to virtual switches, firewalls, subnets, so each VM can talk and be connected with everything that it needs to be connected. And back when virtualization was new, companies had to rethink the fundamental ways that they did business. For people like Maria, these challenges weren't just technical We were standing up this, uh, infrastructure as a service service, and then we were thinking, "Okay, we, we need to charge customers, so they need to come in with a credit card and, you know, enter the credit card." And, and then I remember having meeting after meeting with the billing team. They were used to billing for telephones and cell phones, and they were like, "Okay, how does a customer is gonna sign the PO?" And we're like, "No, you c- there's no time for a PO. This needs to be like an open credit card." By the time that we send them a PO, they sign it, they've used it, abused it, and they're gone. So infrastructure as a service also brought such a huge change for business because you had to think about subscriptions, subscription management, what was the right sizing for a VM. So those are non-technical, sort of business-driven decisions that need to change to adapt to new technologies. That is such a good point that I really haven't thought about in a long time. Uh, Jennifer, have you had any experiences like that? I have. Um, we didn't have a open credit card with it, Maria. It's so funny. Um, but once you get past... Well, okay, so once you are in the position of the folks that set up the hypervisor, and we went, I rambled quite a bit a minute ago about different components of what it makes, you know, to, to create this hypervisor, this platform that people are now, uh, requesting. Your users come to you and say, "Oh, I need a virtual machine. I need a virtual machine." I ne- And before, and you had mentioned it, Emily, you, you buy the server, the physical hardware, you put it in the, into the rack, and then you set everything up, and that's that. Now, they're going to a website and they're pressing a button. And they're getting a virtual machine. How do you control that? How do you say, "Whoa, guys, h- how..." You know, um, a couple episodes ago too I mentioned, um, it was such a sweet story about somebody thinking you could auto scale forever- Oh. ... into infinity. I think I made a joke about a TARDIS. Yeah, same. You can't just keep clicking the button and spinning up new virtual machines- Mm because you're gonna run out of your CPUs. Yeah. You're gonna run, you're gonna run out of the storage. You're gonna run out. And so some folks, yeah, they, it's infrastructure as a service, so you would go and you would give them, "I need a, um, a virtual machine. Here's my $5." So some people did it that way, some people do it tokens. There's a lot of different ways that companies sit down and, okay, also, j- all joking aside, they sit down and they decide, "Within my company, which business units get more compute?" Which maybe are not customer facing or customers facing, so maybe they get a little bit less. You know, there's always- Mm-hmm ... a little bit of a hierarchy, so maybe sometimes our, untor- unfortunately, sometimes our internal users get a little less. Um, you really have to sit down and make some business decisions, as you mentioned, about who gets how much of what. Exactly, 'cause it, it, although it's virtual, it's still physical somewhere. Right. So which means there's still a finite amount that you can scale to, and you can't have, you know, one person in your business or company using all of it, 'cause then, you know, that, that doesn't work. Mm-mm. Nope. But I think, like, that story about, you know, POs and credit cards is so funny, 'cause I think the, I don't know, the timeline of Purchasing things has changed so much in the past decades too. It used to be like, "Okay, I need more server space. Okay, I will purchase a server. Okay, I will set up the server. Okay, I will pay for the server." Now it's like, "No, I need it now." Right now, yeah. "What do you mean I have to wait? No, right now." Yeah. I've got a, I've got- Yes ... I, I know, I keep wanting to, like, say very specific applications and whatnot, but guys, think of anything that you run on your phone when you sit down and you use a web app, whatever. Anyth- any of that is what we're talking about. Like, people need that now. Yeah. They can't wait. They need a update, they need to roll out something new, whatever, but folks can't wait. So we're in a world of, like, everything is as a service. Exactly. Like, we use VMs a lot for, like, demo environments and things- Right 'cause you need to spin it up and show someone something, and I can't wait three to four business weeks for that to happen. Heck, I was at a science fair recently. I was at a science fair, and kids were spinning up virtual machines. Mm-hmm. At a science... I was like, "Where in the heck am I?" You guys, we're in the future. Yeah. So fun. We really are. It just keeps going faster and faster. Speaking of faster, I think we do need to talk about AI. Oh, yes. Because virtualization and AI together is becoming a very interesting story. Virtualization was originally around, um, making it so that you could have more redundancy, you could roll, roll out new systems faster, whatever, you know, define system as you will, um, all of that. And so we're using a lot of CPU, we're using a lot of storage, et cetera. Now, though, in the past, huh, let's say three years- Mm-hmm ... all about GPUs. So the core challenge is that these large language models, LLMs from now on, need a lot of processing power. Yes, and GPUs are expensive, especially these days, and not so easy to obtain. Right. And even when they finally get access to the GPUs, companies simply cannot afford to let them run at 10% capacity. Yeah. I also don't wanna go too much, Emily, well, real quick, I don't wanna go too much into the cost of these things recently. Yeah. They've been skyrocketing, skyrocketing. A-da-da. So if you finally get your hands on some of them, you certainly don't want them sitting around doing nothing. So how do we better utilize the GPUs? Exactly. And Maria explains how that works Cloud providers offer these specialized VMs that have direct access to massive clusters of GPUs, and then the hypervisor is optimized to make sure that there's pass-through the physical hardware directly to the VM so that VM can make the best use of that GPU, minimizing performance loss. Yeah, and that AI is not just running inside of those virtual machines, it's now managing that AI usage. Oh, gosh. You guys, I, I just need to throw out one more bit of jargon just for a sec, 'cause this is the name of a product. There's something called Loki underneath all of this, so everybody needs to go look. Loki, as in- Loki like Loki, or Loki- Yes ... like the Norse god? Like the Norse god. Okay. And I just lo- okay, so one of the reasons that I've been in this industry for so long, one of many, many, many reasons, but one, is definitely the naming conventions. Oh, yeah, you get the most fun names, I think, in tech, for sure. Right. So now we've got all of these different, uh, I'll just keep saying components, that are being used to f- actually manage not just how much GPU is being used by each group within a, a business, or, hey, w- we could even say, like, a family. I know a family that manages a, uh... has their own, uh, hypervisor, where they have their kids go in, and I mentioned the science fair, where they go in and they literally uh, like, "Here's your virtual machine. Here's yours." And so they're managing which hardware those kids are using. Yeah. And then you're going, like, a layer up, kinda, from a user perspective and managing what the, uh, AI, uh, agents and, okay, they have chatbots, but whatever- are using still. No, I think that makes a lot of sense, and that gives us, you know, a real solid use case for VMs 'cause we were talking earlier, like, do you use containers or VMs, and when for which? And so if I'm getting this right, AIs, LLMs use a ton of GPUs. You can use a virtual machine, stick that AI in there, and use that to control how many GPUs it uses so it does not go crazy. Yes. Well done. A- and I, I mean, just to get people's brains going, and they can go out and expand this, is you can also use small language mo- small language models, we haven't mentioned that yet- Mm-hmm ... and have containeri- like, those are containerized. Yeah. There are so many options. And apologies, but if you really can't figure it out, you can actually talk to some of these chatbots, and they will help you figure it out. Mm-hmm. You don't, you're not alone. You're not alone 'cause a lot of people are trying to figure this out right now. Do I run my large language models in their own virtual machine in an environment that has a large amount of GPUs, where I'm sharing those GPUs, or do I run some of these smaller language models, not kidding- SLMs. Um, in container, right? Very specialized, and I'm just gonna go all in 'cause I was just listening to one of our earlier episodes. On the edge. All right, I'm tying it all back together. Anyway, hey, Emily, welcome to AI Ops. Oh, we are in it now, aren't we? I just think that's so cool. Like, I think it's a little understated how fast things have gone. Like, I know people talk about it all the time, but, like, containers are new, VMs are new, AI and LLMs are all new, and they're progressing so, so rapidly fast that finding how they work together has been, you know, a full-time job the last few years. Yes, it has. I'm glad we have people for that. Yeah. We're not on our own. Everybody is in kind of the same boat, which is, I don't know, soothing. And none of them are going. I mean, it's not leaving. You know, it's not going anywhere. We, we need all of it. Mm-hmm. We need to keep buckling down. Um, we now have predictive analytics, generative analytics. Looking back at the historical data- Around how our virtual machines are used, like in the industry that I'm in, has also helped so much in predicting where we're going and what we should do, how we should set up our hypervisors in the future. Mm-hmm. How we should set up our containerized environments in the future. I made a little bit of a joke there about going and talking to your chatbot about it. Not actually joking. No, seriously. You know? Not actually joking. Do that. It helps a lot. Like, I use it sometimes also just to find sources to go read, because there's so much. Yes. There's so much out there, nobody can read it all. Yep. And I even just... Oh, and I just messed up. So I should've said, "Go talk to your chat agent." That's what the cool kids are saying these days. It's all agentified. We haven't said that word enough yet, so we have to say it. It's all agent- It's true. I- like, I'm sure by the time this airs there'll be something new and shiny. There'll be another word for it. Whoever is listening right now, don't worry. We all, there's a new term right now, so go ask the agent. Things move fast. It's okay. You're doing great. You're doing great. And I think, you know, through all that, I think it's safe to say that virtual machines are still an important part of IT infrastructure. As much as things seem to have been replaced by containers, we're not there yet, and I don't think we ever actually truly will be. In the industry that I'm in, there are certainly many, many situations where virtual machines, virtualization, is still powering a lot of the internet and phones that we're using right now. Mm. So it's not, uh, gone away. Maybe in the future we'll talk more about running some of the virtual hypervisors on top of containers, but not yet, not yet. We're not gonna talk about that just yet. We're keeping it one thing at a time for, for now. Maybe we'll, I don't know, do fun combos at some point. In the back of my mind is a lot of the environmental issues also that some of these technologies, that's popped up quite a few times, maybe in folks' minds, but if not, it will from going forward. And a lot of the technologies that we're talking about, virtualization, especially related to AI, we're able to use a lot more of the powerful hardware that already exists. Yeah. And we can optimize that because of things like hypervisor virtualization, virtual machines. So yeah, a little bit of hope. Rather than just kind of brute forcing through it, w- it's introducing a bit more of the, you know, classic reduce, reuse, recycle- Yes ... into, you know, h- using the tech. Oh, I've seen it firsthand. Mm-hmm. So it's some hope. I'm a hope-based creature and there's some hope. Awesome. Okay, so like we usually do, we talked about a lot of things all around virtualization. So let's recap. Virtual machines are essentially kind of renting out space on physical servers that exist elsewhere, much like kinda anything in the cloud. So you don't have to maintain your own set of servers necessarily. You can use just a little space elsewhere, which can be helpful for things like scaling, uh, and things like controlling new tech that is resource-intensive like LLMs or, uh, anything that is CPU or GPU heavy, you can control it within a VM. So they're still distinct in use case from containers. What else did we talk about? Was that, was that it? I'm really boiling it down here. Right? I mean, there was the, the infrastructure as a service, which I love that piece of it. Mm-hmm. But you, you hit that because y- we are, it's about controlling all of these different components of your compute environment, right? And hypervisors to help make sure that you are distributing all of that infrastructure as a serve- service how you want to. Yep. I just get sort of hypnotized by your summaries of these. I'm like, "Wow, she really got it all." It's like I try really hard to pay attention. I know. It's absolutely fantastic. We get lost in the weeds, so, um, trying to make sure we're, we're hitting the highlights here, too. And I think, you know, overall, virtualization has a solid origin story. I think we owe it a lot when it comes to, you know, the bedrock of modern enterprises. Yeah, and far from being replaced by containers or AI, it has adapted to maximize the most powerful hardware on earth. Not bad, right? Not bad at all. All right, now you've heard our thoughts, and it's time for us to hear yours. So hit us up on social media @RedHat and use the hashtag #compilerpodcast. And that does it for this episode of Compiler. This episode was written by Kim Wong. Thank you to our guest, Maria Bracho. Compiler is produced by the team at Red Hat with technical support from Molly Brock. If you liked today's episode, follow and review our show on your platform of choice. See you later!

More about virtualización

La virtualización permite crear varios entornos desde un único sistema de hardware físico.

Publicación en blog

Novedades de Red Hat OpenShift Virtualization 4.21

27 de marzo de 2026
Virtualización
Publicación en blog

Modernice las máquinas virtuales en Google Cloud con Red Hat OpenShift Virtualization

26 de marzo de 2026
Servicios de nube, Virtualización
Podcast original

Command Line Heroes: Cuarta temporada: Una cosa más con Steve Wozniak

Historia de la tecnología
Podcast original

Open Source Hardware | Command Line Heroes

Open source, Historia de la tecnología

Sobre el podcast

Compiler

Do you want to stay on top of tech, but find you’re short on time? Compiler presents perspectives, topics, and insights from the industry—free from jargon and judgment. We want to discover where technology is headed beyond the headlines, and create a place for new IT professionals to learn, grow, and thrive. If you are enjoying the show, let us know, and use #CompilerPodcast to share our episodes.