Learning k8s ep. 3 - The end of the VM

Hinzugefügt: 2026-05-15

[00:00:00]Heat. Heat.

[00:00:21]Heat. Heat. N.

[00:00:59]Heat. Heat. N.

[00:01:44]down.

[00:01:46]I feel All right, everybody. Welcome back. It is Thursday, 11:00 a.m. Eastern, and it is time to talk Kubernetes.

[00:02:23]My name is Marco. I'm the EngineX evangelist at F5, and I will be learning Kubernetes alongside everybody else. Like I've mentioned in the previous two episodes, I have some experience with containeriz containerization.

[00:02:41]That was a word. have some experience with containerization but not a lot of experience with Kubernetes. So I'm learning along the way with everybody else.

[00:02:51]In the previous episodes, we took a look at what my plan is. Cubstronaut five searchs that get you that coveted blue jacket and what the main focus area for learning will be. So I want to make sure that I can help you along the way as I learn and any lessons that I learn are also applicable to your learning and I want us all to succeed in this endeavor.

[00:03:17]Now today is going to be a little bit of a different episode. I'm not going to talk about exams. I'm not going to talk about um exam questions even though I've built that exam prep tool that I'm still iterating on. Obviously it's version 0.0. 002. So very very low. Uh but I I do have a couple individuals who have shown interest to test it out. And yeah, I still am looking for people to to adopt and contribute to that tool. I want to make a learning tool that can be applicable to anybody. And if if you don't like how it looks, how it feels, how it works, you're welcome to fork it and use it for your own purposes in your own learning journey. uh or however you want to use that platform as well. It there's all options are on the table. So I named this session the end of the VM.

[00:04:11]Episode three, the end of the VM. Why am I talking about the end of the VM? What has what are virtual machines? What what do they have to do with Kubernetes?

[00:04:20]What am I talking about really? Well, um I have just spent three days at conferences. I was in Atlanta at Red Hat Summit and I was also attended a Kubernetes community days in Toronto and I talked to probably well I'm going to say over a hundred of people but but it's probably more it's probably more than just over a hundred and with that I have some lessons from the field some anecdotes from the field that I want to share with everybody.

[00:04:55]And I think that we're at a very important inflection point where there's going to be a massive change happening in what we consider as the core enterprise architecture uh for lots lots and lots of businesses.

[00:05:10]And uh let's dive deep into it. F let's talk about virtual machines. First let's talk about what virtual machines are and how they matter. Uh there we go. All right. Um, let me put myself to the side so you can actually see all of this screen that I'm sharing here. So, in general, if we look at the big picture of Kubernetes pods and containers and virtual machines, they're very similar. They're very similar. What we have is in Kubernetes, we have a pod. We can have one or more containers in a pod. Just like in a virtual machine, we can have one or more applications.

[00:05:56]We essentially are isolating those applications, those containers from the underlying physical hardware.

[00:06:04]So the goal of the virtual machine and the goal of the pod is to isolate that application, make it entirely independent from the underlying architecture. Which means in an ideal world, I can take that pod or I can take that virtual machine and I can ship it anywhere. I can drop it onto my desktop.

[00:06:28]I can test it out on my laptop. I can run it on a server in my data center. I can ship it to the cloud and it's going to run without any modification.

[00:06:39]And this is to a certain degree true especially about containers because they're very interoperable.

[00:06:49]Now, is this true about virtual machines?

[00:06:53]Yes and no.

[00:06:55]Why do I say yes and no? Well, a virtual machine is a virtual machine, right?

[00:06:59]It's just a virtual hardware. Uh we've got an operating system inside of that virtual hardware. It's got all its binaries, libraries, and and mind you, these operating systems can be quite different. This could be Linux, this could be Windows, this could be a Mac OS, Mac server, whatever, whatever you want to run on these three different types of virtual machines. Whereas with um with containers, we have a little bit of less choice of the operating system itself.

[00:07:27]So we have to have a kernel uh in the host operating system that supports either the application or if this is a really heavyweight container that contains a different operating system running on the same kernel we can support that. So let's say I'm running some kind of operating system underneath some kind of Linux with kernel version XY Z. Now if I want to run a different distribution of operating system inside of a container I can as long as it supports the XY Z kernel. Is that important? Well that is important to differentiate between uh especially capabilities of containerized platforms and capabilities of virtual machine platforms. So I have variety. I have choice. Now is variety and is choice something that's good?

[00:08:19]A lot of times we see that when we have a certain unified environment, when we have hundreds, thousands, tens of thousands or even more systems, what enterprises tend to do is they tend to unify on a certain operating system on a certain platform.

[00:08:41]So what's going to happen in hypervisors is a lot of the times this guest operating system nowadays probably 90% of what enterprises run is going to be Linux. It's going to be a certain Linux at a certain kernel. It's typically even a certain distribution from a certain vendor that offers support to that company.

[00:09:03]So it's going to be uh you know YUbuntu or Kubernetes or sorry not Kubernetes or Red Hat or some other a type of Linux operating system that actually has support attached to it.

[00:09:17]So what you're going to see is that rather than seeing a lot of diversity in the guest operating systems on a specific hypervisor, you're going to actually see that there are pretty much the same machines running on top of that operating system guest operating systems. They're all they're all roughly the same. So we got a lot of duplication here. A lot of duplication happens.

[00:09:45]Okay. So if we have the same applications and roughly the same operating systems, then what is the major difference between your containers and your virtualization platforms? Well, it's it's the control plane. Uh the hypervisor lets VMs be run on top of that specific hypervisor. Whereas with Kubernetes, the control plane allows us to run the pods. Now, here is where the big difference with cross environment support comes from.

[00:10:23]Kubernetes is designed as a cloudnative platform.

[00:10:28]It's ran by the cloudnative computing foundation and the purpose of Kubernetes is it has to be agnostic on what it runs on completely agnostic about hardware agnostic about what kind of platform there is and and no vendor lock in essentially to to the underlying component.

[00:10:49]So you're on Kubernetes again on your laptop on your server or on the cloud.

[00:10:54]It's the same pods very very interoperable. Now what happens with the virtual machines is you get locked into the hypervisor. You can't just drop virtual machines from VMware or from HyperV or from KVM or from Zen or whatever hypervisor you're running. You can't just run those on another hypervisor. Interoperability between hypervisors is severely limited.

[00:11:21]Severely limited interoperability. There there are some unified standards for unified virtual machines, but really nobody uses that. If you're running VMware, you're going to be running VMDKs, you're going to be running a VMware machine configurations. If you're running HyperV, you're going to be running VDKs. Yeah, configurations for HyperV. There is no interoperability between hypervisors. And this is where the big move from virtual machines to the cloud is actually going to happen.

[00:11:56]Now why am I saying this? uh like I said I spent three days at conferences uh talking to dozens over hundred people probably and while there was talk of AI there was a lot of talk of AI the biggest conversation was we want to move off of our virtualization platform we want to have virtual machines that actually do what was promised about virtual machines things like cloud bursting. We want to run them at the edge. We want to run them inside of a data center. We want to run them anywhere we can uh in the platform itself. That doesn't matter what the platform is, doesn't matter where we're running. And and this is the thing that the virtual machine and especially the hypervisors that are ran by different vendors, companies, this is where there's sort of a disconnect between what the customers want and what the hypervisor actually offers. I'm not saying that there isn't a capability that allows you to move your virtual machines to the cloud. There definitely is there the there capability exists depending on the vendor. There's going to be different support in different cloud providers. Um but it gets super complicated. It gets super complicated or requires a special license and that license can be very expensive and this is where customers are finding uh that right now having virtual machines is more of a burden than a benefit.

[00:13:46]having virtual machines is not really turning out to be that promise of flexibility, the promise of scalability, the promise of quick spin-ups and things like that. Whereas on the other hand, containers spin up in seconds. They're super lightweight. Um they're super portable to any platform, to any cloud, to any physical box that has a processor and some memory essentially and perhaps even some storage if need be.

[00:14:15]um we we can obviously cut out the the overhead of the guest operating system.

[00:14:21]How how much memory or CPU do we need to run our application? At the smaller scale, the applications need megabytes, a few cycles on a CPU. Whereas the guest operating system, just to spin up the guest operating system, you're using gigabytes of memory. you're using one or two CPUs for the operating system to actually keep itself alive and and do all the things that an operating system needs to do. So what we're getting is we're getting much higher density because we're cutting out all of these duplications of operating systems that are running across the board. Now there are hypervisors with memory optimizations where if you're running one operating system and you're duplicating it 10 times, there is some shared memory pool in there. you can turn that feature on or off. So again, perhaps for security um you want to have your machines entirely isolated from other machines. You don't want to share anything. Each uh VM is more secure technically. I I would beg to differ that there is security and I'm going to get into security uh in a second, but for now containers, they spin up fast. They're super high density. Uh they're super portable across environments. Um they can be f really efficient like super super efficient. Better h at at resource utilization better at spinning up scaling out scaling in all happens in seconds. Whereas with virtual machines, minutes, maybe tens of seconds if you have super highly optimized Linux operating systems that can actually boot up in tens of seconds.

[00:16:02]I know one of my machines that I optimize specifically for one task, I can get it to boot up in about 18 seconds from powering on a machine to actually getting a command line in in Linux, but it doesn't load really anything.

[00:16:18]Everything else is loaded um after. So really all the applications on that machine are ready within minutes.

[00:16:28]Yes, the operating system takes 80 seconds to boot, but the containers applications, they spin up afterwards.

[00:16:37]So, we're looking at minutes for startup for virtual machines. And if you have a an application where there's very spiky traffic, it's very difficult to predict that traffic, it's very difficult to satisfy that traffic with virtual machines.

[00:16:53]So, what is the real benefit of the virtual machine? Well, technically um this this is where the difference starts, right? Pods, you can think of them as virtual machines. Containers, you can think of them as apps inside of those virtual machines.

[00:17:10]The difference starts at where we're kind of like here in the green. It's the host operating system or a bare metal operating system for that hypervisor itself.

[00:17:22]Host operating system, we're going to need Linux. We're going to need Windows to run those containers.

[00:17:28]And we know that securing an operating system, giving it access to giving a user or an admin or somebody who has access to the operating system, we're technically allowing them to see all of these pods.

[00:17:42]We're technically allowing them to see the whole uh yeah cluster space. So the real attack vector for containerization of any kind would be getting onto one of those host host operating systems and being able to see a lot of containers, a lot of applications.

[00:18:02]We're going to hear from everybody who's confident about hypervisor being more secure say well yeah the vector of attack for a bare metal hypervisor that I can typically also get into. So I can either SSH into it or get into it via some kind of API process.

[00:18:25]The bare metal hypervisor, the common belief is, oh, you know what? It's more secure because there's just not many known attack vectors.

[00:18:40]And this this is the this is the message that I heard from well dozens of people.

[00:18:45]At least a dozen people said, "Well, yeah, but the the hypervisor is is so secure that nobody knows how to exploit it." But then when nobody knows how to exploit it, the bare metal hypervisor, nobody even knows how to protect it either. So a lot of times these bare metal hypervisors don't have any kind of security tools on them. A lot of times it's because there's no security tools that fit onto them and other times it's because the sort of perception of hypervisors as being super safe is probably a little overconfident.

[00:19:26]Not protecting an attack vector because it's deemed to be oh you know this is safe safest houses that that is that is not a good approach. So these anecdotes that I'm hearing from individuals saying, "Oh yeah, we prefer the hypervisor because it's more secure than the operating system that runs containerization."

[00:19:47]That could be a little bit of a obscurity security by obscurity fallacy here because you can protect your operating system and there are known ways to audit, protect, um, detect, prevent any kind of shenanigans going on in your operating systems. Whereas on the other hand, there are very few ways I mean it's kind of it can be kind of difficult to actually have the bare metal operating system built up from a security standpoint to the point that you can build up your host operating system.

[00:20:29]So to that respect, it's actually easier and you can probably achieve a better compliance posture with a traditional standard operating system than you can in some situations with a bare metal operating uh bare metal hypervisor.

[00:20:48]And yeah, at the end of the day, we still have the physical server, we have the infrastructure, and here it's physical access, you know, just not letting people grab your servers and and take them out of there. So, what is the future for virtual machines? Why are why are we still using virtual machines?

[00:21:09]Well, you know, they're still they're still good. They still do things. they uh they still are there still are situations where we want to pack different things together. Um we want the monolith because of the capability of the monolith. Let's say um I've got a a big video I'm sharing I'm sharing a big video. I'm ingesting it here with pod C. Uh pod C inspects the video and it's you know tens of gigabytes. Pod C inspects the video and then passes it over to container A in pod one here. Uh which then passes it over to D which then passes it over to B which then passes over to E. There's a lot of network overhead happening here. Now obviously we can save that we can use a shared volume. We can store the video on a shared volume but there still might be retrieval from to and from the shared volume that uh offers an overhead. So if I if I package all those applications into one virtual machine and I have this scenario where I'm dropping uh the video and I got the pipeline in that virtual machine that's going to get loaded into memory once and we're good.

[00:22:16]Now that is a valid scenario but that can be done in a single container. You can have a complex pipeline running in a single container. It doesn't need to be designed to be super super microsered.

[00:22:31]You can you can have these microliths that do a certain specific task that may do perhaps many certain specific tasks uh efficiently in memory on one unit of of processing data. Right? Think of that video as the processing data that is coming in.

[00:22:49]Technically, like we said before at the beginning, um you can have a full guest operating system as long as it shares the kernel of the host operating system.

[00:22:59]And while yes while there is some limitation of what kind of operating system is supported that interil interoperability between operating systems is growing is getting better. Um you've got the Windows subsystem for Linux that as an example if you run a Windows host operating system you can run pretty much any version of Linux that supports WSL.

[00:23:23]Um, am I advocating for you to run Windows servers for your Docker uh just so that you can run uh WSL supported container containerized operating systems? No, not necessarily. But it's an example of how interoperability that can be very easily used by everybody uh exists out there and we have that.

[00:23:47]So, do we need virtual machines for security?

[00:23:52]I'm not going to be I'm not going to be an advocate of that anymore. I used to say, "Yeah, sure. You can bring your security up. You can bring your security posture up um at the guest operating system level. You can encrypt the guest operating system. You can run tools for security inside of the guest operating system. But you can do that in a container as well. You can encrypt your volumes in a container. You can pull a encryption key from somewhere else from external storage and you know have ephemeral keys that you can use only once like your one-time password tokens for example.

[00:24:29]Um no difference it's no real difference between a container or pod and a VM at this point from from the respect of security. Uh and then the the last argument the last argument was but well what about legacy applications?

[00:24:47]What about legacy applications? Well, what's going to happen to legacy applications is number one, they're not going to stick around forever.

[00:24:56]And number two, they're probably going to go away at some point or be relegated to maybe some very niche virtualization platforms that are going to still support virtual machines.

[00:25:14]I believe my personal belief right now is that Kubernetes is the future of virtual machines.

[00:25:24]Yes. While you're going to be keeping some uh you know some hypervisors there somewhere in a data center to run your legacy platforms al as long as they're required or maybe you you port those legacy applications up into the cloud onto virtual machines in the cloud. Probably your best scenario at this point so that at some point you can just quickly archive them and drop them into the archive tier of the cheapest object storage solution you can find.

[00:25:55]your core business is going to run on Kubernetes.

[00:25:59]Okay, why Kubernetes? Why not Open Shift? Why not just Docker containers?

[00:26:04]Docker Swarm does things, right? You can do that with Docker, why why not the cloud versions of of um containerization like why not ECS?

[00:26:17]Why not, you know, why not the capabilities that exist that cloud vendors have?

[00:26:24]Well, because what happens in an industry is there is something called industry standards. And your industry standard for virtualization up until this point in the majority of businesses, if you've been in it for a while, you know your industry standard is VMware.

[00:26:43]Why are people not using hypervisors that are not VMware as much?

[00:26:50]Well, number one, VMware was exceptionally good at the beginning of its sort of first 15 years of its existence. It was exceptionally good. It actually outperformed a lot of hypervisors. I personally was a VMware uh a VMware VCP up until version five, I think back in the day. And when when we were testing capabilities, when we were testing interoperability, VMware came out on top. Uh it just it just worked. It was very stable. It was solid. And then also the community that supports that platform is immense. It's huge. You're probably gonna find 10 times more people who are familiar with VMware than any other hypervisor. I'm not saying that other hypervisors are not in use. There's tons of HyperV.

[00:27:47]If you have everything running on Windows and if you're connecting your platforms to Azure cloud, then HyperV is probably the best supported most flexible virtualization platform that you can use for sure, but it's not the industry standard, right? And when you have other options, sure they exist out there, but then one thing becomes the industry standard. And this is where I personally believe Kubernetes has the probably the best the lead in becoming the industry standard for the next several decades for how we run applications. how we run applications both on premises and how we run applications in the cloud and then um the question about on premises or cloud uh comes comes up um what's what's happening with cloud what's what's happening why why is there an exodus from the cloud there isn't an exodus from the cloud that number one let's dispel that myth 95% of spend in IT globally about $5 trillion is spent on it globally. 95% of that is spent off of cloud.

[00:29:11]Cloud is a you know 300 $350 billion billion dollar business across all the cloud providers. So yeah um you know 90% okay let's say 90% of that spend is off cloud. Even if 90% of that IT spend is off cloud, that means that there is a a lot a very very large majority of on premises applications. And this is what was confirmed to me with the discussions that I had.

[00:29:46]A lot of companies are regulated. A lot of businesses are regulated. They run air gapped systems to if you're running an airgapped system, you cannot be using cloud because you can't you can't physically access it. You cannot go to the cloud data center and access your application if it's airgapped, if it's not connected to the network. So those will always stay in data centers. That's number one, air gaps. Number two, um it's not that cloud adoption isn't happening. There's there's cloud adoption happening and there's also the evolution of cloud services that's happening. A lot of companies lifted and shifted figured out, hey, you know what happens when you lift and shift a thousand virtual machines? Your bill goes up. You're not saving money anymore because that that was never supposed to be how cloud should be used. You shouldn't be lifting and shifting tens of thousands of virtual machines from your old platform. You should be modernizing. You should be implementing containerization because it's way more efficient, especially on cloud. You're paying in very small units when you run containers and you can ephemerally spin them up and down. Whereas when you spin up virtual machines, most of those virtual machines in the cloud, they run 247 even if they're not doing anything.

[00:31:07]So a company had a thousand virtual machines on their local cluster in their local data center and they pushed them into the cloud and they kept them running 24/7 just like they did on premises. No change in the evolution of their processes, no change in how their applications run and virtual machines again were not proving to get that would give that benefit back that that flexibility that scalability that was promised.

[00:31:36]and bills increased. And a lot of businesses when they figured that out, they said, "Oh, you know what? We don't have the technical acumen required to modernize this application.

[00:31:48]If we're going to run virtual machines in the cloud, it's going to be super inefficient. Let's get off cloud." And they moved the virtual machines back into data centers because that made sense for them.

[00:32:01]You can only win with cloud when you modernize your application. You can only win with cloud when you use it as an ephemeral resource. A resource where you spin things up when you need them and you shut them down when you don't need them. And regardless whether it's hardware cloud, whether you're running multi cloud architectures, the way to get that benefit out of cloud is to run Kubernetes, is to run the platform like Kubernetes. Number one, it's positioned, like I said before, it's positioned to be the platform to run applications because just like VMware was 15 years ago, it's got a huge community that supports Kubernetes itself.

[00:32:42]It's got a huge there's there's a huge drive from the CNCF itself and from all the cloud vendors to support CNCF capabilities in a native manner in a way where it is interoperable out of the box. It's not a hack between how do I get my VMware virtual machine into the cloud because the cloud supports a different thing that they've built themselves. No, it's it's all supported.

[00:33:10]It's all you take the pod and you just target it to the cloud. You target it to the data center.

[00:33:17]Why not Docker? Why not Docker Swarm?

[00:33:19]Why why not Open Shift? Why why not all of these other capabilities that exist out there?

[00:33:25]Cloud Foundry, I don't know. Think think of of your favorite containerization platform because of the industry standard. The industry has rallied around Kubernetes as the standard for running containers.

[00:33:40]uh it it is heavyweight. It's got a a heavyweight control plane. It's got it's designed to run at massive scale and there are scenarios where Docker can be leaner where you know the the flexibility that Open Shift brings to the table is still better for some businesses. the specifics about for developing applications that cloud foundry links brings to the table. Those are still valid points, but they're not going to be massively adopted like Kubernetes is being adopted and will still continue to be adopted. And and this is where this is where I think that we have that edge with Kubernetes o over all the rest of the containerization platforms. the drive by the community, the adoption, the the massive amounts of people who are who are supporting Kubernetes platforms and businesses will realize that hey this is a enterprisegrade well supported um platform where where the workforce is growing for it a lot more people are are learning Kubernetes and I think if you want to future proof your career it's not just Linux or just cloud or just Kubernetes. It's it's all of those together.

[00:35:05]It's how do I run my containers?

[00:35:09]How do I develop for those containers?

[00:35:11]How do I build pipelines for those containers? And then I can run them on Linux and then I can run them on cloud and then I can run them on all of those platforms. Sounds like a lot, but that's that's how I feel about it.

[00:35:26]All right.

[00:35:28]That that is that is the the lesson that I think I've learned. So what about the legacy applications? Did I answer that question? Well, they're going to stay on on virtualization. They're they're definitely going to stay on virtualization. They're stay they're staying on some kind of platform somewhere. They're analogous to how mainframes are today. Mainframes are still there. Mainframes are still there.

[00:35:52]They're they haven't gone away.

[00:35:55]IBM is still building their mainframe platform, still selling it. Today, you can buy a brand new mainframe. If you want to run your applications from the 1970s on on mainframes, you can still buy them. Does everybody buy them? No.

[00:36:11]There was an interesting an interesting um research done a recent rec recent article um that talked about how it it actually could be uh more efficient to run virtual machines on mainframes nowadays with the lack of hardware with the lack of um capabilities if you have a certain scale of VMs the the analysis showed that it actually makes sense to go for the mainframe platform that the Z series can actually support. I think it was something like 700 Linux virtual machines. And then there's the Sephur AI accelerator that IBM has built for the main frames so that you can actually run workloads like large language models on on those accelerators and and they're they're available. You can actually buy them not like the not like the GPUs that you can't buy.

[00:37:02]All right. So Kubernetes containers this is the way to go. This is what is going to promise. This is going to deliver on the promise of run your application anywhere. Run it on the cloud. Run it on any platform. Move from one cloud to another. No vendor lock in. Use the same automation. Use your you know Kubernetes orchestration documents not some vendor specific orchestration documents.

[00:37:33]That's that's the main idea here. uh we want to develop our infrastructure as code. So when we're developing our infrastructure as code, we kind of have to decide on a specific language that we're going to use for that orchestration. Um is that language going to be cloudspecific or is that language going to be vendor dependent? No. In when you're running Kubernetes, all of those orchestration engines and pick your favorite orchestration engine here.

[00:37:58]I'm not going to even mention any of them because there are polarizing views between orchestration engines supported on Kubernetes. So, pick your favorite Kubernetes orchestration engine.

[00:38:12]uh and it runs on any Kubernetes platform whether it's a vendor specific Kubernetes platform that integrates some vendor features like EKS on AWS for example or whether it's just plain vanilla uh Kubernetes or it's the Kubernetes that's in GCP which essentially is the same platform itself it's going to work across all of these platforms so your disaster recovery becomes much easier your migrations become easier you have no really you don't have any vendor lock in. Uh you're running on open source solutions which it's the best ever and um yeah you get all the benefits of containerization.

[00:38:53]Yeah, some legacy applications, hey, you're going to have to keep them on VMs for some niche things and then you're going to get rid of them at some point or pack it into a container image. build a container image from that operating system and squish it onto Kubernetes as a single uh single pod that restarts if it fails.

[00:39:17]Yeah, there there are pathways to achieving that as well that exists.

[00:39:23]Um and then one final point I want to make about this whole rant that I'm having here.

[00:39:30]There is something happening in local compute as well. Edge computing is becoming a really really important part of computing because computing is becoming so powerful.

[00:39:47]Everything that we build nowadays is extremely power dense. Um, and what I saw at the Red Hat um, Expo was one of the vendors was showing one of their newest generation servers, a three-unit server. Um, two processors in there, a little bit of memory. The two processors had 386 cores, physical compute cores, and the next generation is going to have 512. Um, and you know, several terabytes of data.

[00:40:20]That device there was fitted with four terabytes of data.

[00:40:25]Sorry, memory, not data, four terabytes of memory. I mean, and you can squeeze more uh depending on obviously the support of the the CPU itself. So we're looking at hundreds of cores and terabytes of memory that are in at each three units of of your rack.

[00:40:44]So repurpose your your essentially communications rack to have some local compute capability.

[00:40:52]Every every office building has some switches they need because you got some wires going into desks where computers are plugged in and and there's got to be three units in that rack. If it's a 19inch rack, there's going to be three units available to drop one of those servers in. And if you're going to be accelerating your applications with a local copy, I'm not saying run everything on premises. I'm not saying convert all of your network racks into data centers. I'm saying add some capacity to the edge where you can run a copy of your application that runs at local network latency that runs at submillisecond response times.

[00:41:35]And if you want to do that, then you got to span your architecture from the cloud, from your data center, from everywhere into the edge.

[00:41:46]When you do that, a platform like Kubernetes is going to have the most benefit. We've already established it's super lightweight. It's super efficient. You don't need to put the 300 core servers in there. You can just put a one unit server. one unit server will probably have at least you know 64 maybe 128 cores in there nowadays.

[00:42:10]That that's that's the beauty of this capability. This this massive increase in how much compute capacity we have available on our local systems is that we can actually do that promise of just computing is everywhere. computing sits in every rack and has the capability to support your entire enterprise and and when you have obviously that local server, you know, fails because something happens, then you have a copy of that application or more pods um of that service running inside of your data center, running in the cloud, running somewhere uh somewhere else where they should be.

[00:42:57]And this this is what Kubernetes brings to the table. I'm not saying it can't be done with other platforms. I'm saying the simplest way and the most enterprise ready and the most supported way of doing both cloud computing, data center, hybrid computing and edge computing is going to be by utilizing Kubernetes as your platform.

[00:43:16]You're not going to achieve that on your virtualization platforms unless you're vendor locked in into one specific vendor. I think that the only real way to be entirely vendor neutral in this scenarios is going to be Kubernetes.

[00:43:34]All right. Um this is where I think I'm going to conclude my point and maybe the end of the VM is the uh clickbait title that I I put on this video, but you can kind of see where I'm going at. And and this is coming from anecdotes of individuals that I spoke to that that I talked to at conferences that say, "Hey, we're seeing this same thing. We we want to we actually want to get rid of VMs. We want to get rid of that overhead. We want to get rid of how we've done things. We want to get to containerization." And the path is still long for a lot of businesses, a lot of teams are still looking at that as the long-term future in the next 10 years.

[00:44:26]But if you want to be um you know, if you want to have a career in computing, you need to be looking at 10 years from now.

[00:44:36]And 10 years from now, two things are certain. Um you're going to need to make yourself useful to AI.

[00:44:44]think of what you can do for AI, not what AI can do for you. Um, and you're going to need to be able to think in, you know, future sort of project what's going to be happening in the future. I think that our problem solving, our ability to think outside the box is something that's going to be very difficult to replicate with AI.

[00:45:09]Um, but yeah, it's still going to be a big part of of business. It's still going to be a big part of platforms. So, yeah, I'll leave you with that thought. How can you make yourself more useful to AI?

[00:45:28]All right, everybody. Um, I'm going to check the chat if there's any questions or anything that anybody wants to add.

[00:45:36]And if there's uh nothing that's being added, I'm going to say thank you so much for watching this one.

[00:45:44]Uh I hope you find this style of video a little bit a little bit of a break from the norm. I hope you find it interesting. Um I'm going to also say hey before we go uh I want to remind everybody that I have started working on the cubster exam prep tool. Uh so essentially now it KCSA is supported KCSA supported uh take a look at this download it try it out um I've only have 40 questions right now for the exam style questions I only have 10 hands-on lab you can run them on mini cube you can run them on any proof of concept Kubernetes don't run them on your production cluster um yeah so those challenges are there play around with them leave comments.

[00:46:35]Um, leave bugs. If you're going to branch out, branch fork, you know, start this repo. I want I want to build something really cool. And as I add more questions, as I add more content, I'm going to want to see anybody, whether you are learning Kubernetes or whether you are already a Kubernetes expert, I want to see you uh tell me what you think about this tool.

[00:47:01]And I want to make sure that this tool is available to everybody um to take, improve, make better. if you're making it better, I'd love to see your contributions be contributed back to um back to the uh main repo so we can build something really really cool.

[00:47:21]Thank you so much for um for taking a look at this one. Uh yeah, here's the link to the here's the link to the uh GitHub for this one.

[00:47:34]Thank you. Thank you.

[00:47:37]Thank you. Um, yeah, it the videos are recorded on these spec these these same links where you're watching from right now. And I will see you next week. Same place, same time. Enjoy everybody.

[00:47:53]I'm going to end this one. Bye.