Introduction to virtualization - vSphere教程

- [Rick] In this video, I'll be providing a basic introduction to virtualization, and we'll learn what a virtual machine is, what an ESXi host is, and how do I provide resources to my virtual machines. So let's start with the very most basic concept. What is a virtual machine? And a virtual machine is, in many ways, very similar to a physical computer. A virtual machine is going to have an operating system installed on it just like any other computer. That operating system could be Windows, it could be Linux, and the operating system is not going to be aware that it's running on a virtual machine. As far as the operating system is concerned, it's running on a computer just like any other computer. The operating system will have no awareness that it's running on a virtual machine. But one of the big differences between a VM and a physical computer is the virtual machine doesn't have its own dedicated physical resources. So a virtual machine will run on a physical server, called the host, with many other virtual machines, and all of the virtual machines on that host will share the underlying physical resources of that host, and this is the first big benefit of virtualization. We can run many virtual machines on a single physical system, and each virtual machine will have its own independent operating system that is unaware of all of the other virtual machines running on that host. And those virtual machines are sharing a common set of physical hardware, the hardware of the host, so things like the host's memory, CPU, network connections, storage adapters. All of the physical resources of the ESXi host are being shared by the virtual machines running on it. And that's really what the ESXi host is. It's the physical server upon which all of these virtual machines will run. And a virtual machine is almost like software running on a computer. So I can open up my laptop, and I can run Office, and I can run Google Chrome, and I can run all these different applications, and they don't necessarily interact with one another, and they don't know that each other are running, but they're all using the underlying hardware of my computer. Well that's sort of like my virtual machines. I can run virtual machines on this ESXi host. They're not aware of each other, but they're all running on the underlying hardware of the host. And the VMs themselves may consume varying amounts of resources. So some VMs may consume a lot of memory or a lot of CPU. Some VMs might run databases, so they might create a lot of storage traffic. All my VMs can have different resource needs. And so, we're referring to our host as an ESXi host. Well really, what that means is we've got a physical host, and the physical host itself is the actual physical server that ESXi is going to run on. ESXi is the hypervisor that's running on top of that physical server. So basically, the way it works is I will purchase a physical server, and that's my host, and instead of installing a normal operating system on that physical server, I'll install a hypervisor, and that's called ESXi. And the hypervisor is the software that allows multiple virtual machines to run simultaneously. So now I've used this word hypervisor a few times, and that's what the ESXi host is. It's an example of something we call a Type 1 hypervisor. And the hypervisor adds something we call a layer of abstraction, and the layer of abstraction is basically this software that's running between the virtual machines and the physical resources that those VMs are going to consume. So let's use a simple analogy to understand what a layer of abstraction is. Let's say that I'm sitting on the couch, and I have a bowl of M&M's, and I see my sons walking into the room. And I take something like a napkin or a paper towel, and I cover up that bowl of M&M's. Well I've created a layer of abstraction. Now they can't see the underlying physical resources inside of that bowl. And now I have the ability to distribute those physical resources however I see fit. So if there's 20 M&M's, maybe I give 5 to each kid, and I keep 10 for myself, and they will be happy. They'll be satisfied with getting only five because they don't know what's underneath the surface. They don't know that there's going to be 10 M&M's left in that bowl. That's kind of like what the hypervisor does. So the virtual machines in this diagram are sort of like the children in my analogy. They are the ones who want to consume the resources. They are the ones who want to eat the M&M's or consume the memory and CPU and storage and network throughput. The hypervisor is the layer of abstraction that sits between those resource consumers and the resources that they want to consume and intelligently allocates certain amounts of those resources for those VMs. So for example, this particular virtual machine may need memory, and we may choose to grant this virtual machine four gigs of memory. Whereas this other VM might need less memory, and maybe we'll only grant this virtual machine two gigs of memory. And I can start to create these resource allocations based on the anticipated needs of the virtual machines and based on the priority of the virtual machine. And that's really what virtualization is. The VMs aren't seeing the actual physical hardware. They're seeing a version of it that is virtualized through software. They're seeing virtual memory and virtual CPUs. They have virtual network adapters. That's the job of the hypervisor is to virtualize those resources. And one of the most important resources that we're going to provide access to is our CPUs. So here you can see in our little diagram, we've got four CPUs, and I've got a total of 12 virtual machines shown here in the diagram, as well. And so, we've actually got less CPUs than we have virtual machines, and those virtual machines themselves may be configured with one CPU or two virtual CPUs or four virtual CPUs. The number of virtual CPUs that I give to each VM can be different. But the reality is I have more virtual machines than I have CPUs. And so, there's no way that I can give one CPU to every VM. And this is a concept that we refer to as oversubscription. And anytime we talk about the concept of oversubscription, I like to use an analogy, a restaurant analogy, and you'll find that most of my analogies have to do with food. (laughs lightly) So here's my restaurant, and let's assume that we're building a restaurant. Right, so I'm going to build this restaurant. I expect 200 people to come through my door every single day. Now that being said, at any given moment, I'm probably not going to have 200 people there. You know, maybe I'll have 50 people at breakfast time and 50 people at lunchtime, and maybe like 40 people will come for an early dinner, and 30 people will come around 6:00 PM, and 20 people will come around 8:00 PM for a late dinner. And so, even though I know I'm going to have 200 people per day, if I can just build a restaurant that'll seat 50 people, that's going to be plenty. That's what we call oversubscription. I'm not going to build all of the resources that I could possibly need because it's unlikely that I'm going to need them all at once. And so, if we think about this scenario where we have 12 virtual machines but only four physical CPUs, that's okay because odds are all 12 of my VMs aren't going to need 100% CPU 100% of the time. And so, while some VMs may be really busy, other VMs may be not so busy, and I can get by and provide good performance, even though I'm sharing 4 CPUs across 12 virtual machines. And that's the concept of oversubscription. We want to make the most efficient use that we possibly can out of the physical resources that we own. And that's one of the things that the ESXi host does really well. So let's take a moment to review some of the key concepts that we learned during this video. We learned that a virtual machine has an operating system, and it has virtual hardware. So it's not going to have actual dedicated physical hardware, but it's going to run on a hypervisor, along with many other virtual machines, and in the case of vSphere,