[MUSIC] Now let's talk about virtualizing power systems and this thing called power VM. Now when you think of virtualization, you probably think of this a computer, the big box being divided into little boxes, each in some way sharing the resources of the big Bucks. Now, if you don't know what virtualization is there, you have it seriously that's really pretty much it, right? Of course the detail is where it gets more interesting. Right now the x86 system is generally virtualized using a product called VMware. The resulting little boxes are called VMs or virtual machines. And as will soon see run their operating systems. And as we see here, the high level concepts of virtualization on power systems are identical. There are some terminology changes, and of course the virtualizing product is different, it's called power VM. And you see that the VMs are now called logical partitions or L parse or just partitions. Now use all three of those terms, logical partitions, partitions, L parse partitions and even VMs throughout most of the courses. I won't attempt to argue using VM versus logical partition, they both have their merits in terms of accuracy. Arguably though, logical partition is a bit more descriptive. Well I must admit I favor the sound of VM, so let's just say I'm going to stay neutral in that debate. Now, taking a closer look at the logical partition, you see that each can be viewed separate from the big box as an independent operating environment, mostly independent anyway, okay? Specifically, a logical partition is implemented as an abstraction of the system hardware resources processor, memory and IO. Is meant to support multi kernel environments on the same physical system. That is, different kernels in different partitions on the same machine. The partitions are logical because the division of resources is virtual and not along physical boundaries. However, there are limitations. These limitations are defined in a set of configuration rules. The system uses firmware. This is the critical part of power VM to allocate resources to partitions and to manage the access to these resources. This firmware is underlying software that is running on a system independent from any of the operating systems running in the individual partitions. And although like I said there are configuration rules, the granularity of the units of resources that can be allocated to partitions is flexible. For example, you can add just a small amount of memory if that's all that is needed without a dependency on the size of the memory cards. Also you can add memory without having to add more processors or IO slots that might not be needed. So in the end, you can think of the logical partitions operating system environment in the same way you see it on a bare metal system. You have a console, right? And then you have the storage in the data areas that hold the operating system, the application programs, and in some cases the applications data and then application system logs. The operating systems are in dependently tunable in each logical partition on the server. And of course each logical partition has its own network and time and date settings. You could see the logical partition is truly an independent operating environment even though it is sharing hardware with other logical partitions. Also understand that while a logical partition is created in instantiate it on the server, it can go through state changes in the course of its life span. In addition to the obvious power on power off state changes, similar to booting and shutting down the operating system, but not exactly the same. The logical partition could be relocated to another server due to failures, that's called VM relocation or live partition mobility. Or for maintenance or performance reasons, it can undergo a dynamic resource allocation or deallocation change, also known as DLPAR. And on some models it can be suspended and then later resumed. Now taking the resource allocation story a little bit further, let's look at this in another way. You can see the resources of the big box that they are allocated and managed by the hypervisor into each of the logical partitions, which are shown here is boxes surrounded by dotted lines all within the larger blue shaded big box. The mix of resources is controlled by configuration settings and can be changed without disrupting the partitions operations. The assignment of the IO slots in the big box necessarily assigns the adapter that's in that IO slot and likewise any of its attached or peripheral devices being assigned as well, okay? The list of possible IO adapters varies by model, but you can assume that there's a supported adapter for all common types of IO for all our systems. And although I'll cover this in detail later, the IO concept is extended to include virtual IO via the assignment of virtual IO adapters. And shared resources as would be expected, include things like electrical power, cooling fans, core memory, and of course the chatse. [MUSIC]
