Companies of all sizes worldwide are looking to virtualization as a seemingly game-changing scenario. Server consolidation, energy efficiency, simpler management and deployment, and increased capacity are all tangible benefits to be gained from a move to virtual servers and virtually hosted services. Microsoft has seen the light and is here to help with Hyper-V (formerly known by its codename, "Viridian," or by the previous brand name, Windows Server Virtualization), which, according to the company, "is a next-generation Hypervisor-based virtualization platform integrated with the operating system that allows you to dynamically add physical and virtual resources."
You might know about virtualization in general, but you might not be familiar with what the buzz is about. Let's break Hyper-V down in this chapter: how it works, its major benefits, and when you can expect to be able to deploy this feature in production environments. I'll also present a virtualization strategy outline as you consider how to integrate Hyper-V or other virtualization technologies into your workflow.
The idea behind this chapter is to give you a fresh look at Hyper-V in its prerelease state as close to the release of Windows Server 2008 as possible. Some of the information and specific procedures in this chapter may change before Hyper-V is released to the public, but the architecture information and the general steps to manage Hyper-V should be very similar.
To understand Hyper-V, consider its three main components: the hypervisor, the virtualization stack, and the new virtualized I/O model. The Windows hypervisor basically acts to create the different "partitions" that each virtualized instance of code will run within. The virtualization stack and the input/output (I/O) components provide interactivity with Windows itself and the various partitions that are created.
All three of these components work in tandem. Using servers with processors equipped with Intel VT or AMD-V enabled technology, Hyper-V interacts with Hypervisor, which is a very small layer of software that is present directly on the processor. This software offers hooks into the management of processes and threads on the processor that the host operating system can use to efficiently manage multiple virtual machines, and multiple virtual operating systems, running on a single physical processor. Since there are no third-party software products or drivers to install, you get nearly guaranteed compatibility without the difficult problems that software bugs can introduce into your system.
Along with efficient process management, you can hot-add resources to the machine hosting your virtualized services. From processors to memory to network cards to additional storage media, you can add these devices to the machine without needing to bring down any services and interrupt user sessions. You can also host 64-bit guest sessions, which is a big boon to organizations moving toward adoption of 64-bit software. You can virtualize your migration, save money on deployment costs, and then assess how many physical machines you'll need when you finish your migration.
Part of the idea behind virtualization is not only to eliminate machine duplication and save on costs, but also to ensure that services are more available than they otherwise would be on unvirtualized servers. In that context, Hyper-V includes support for clustering across multiple guests. Additionally, you can cluster multiple physical machines running the Hyper-V component, so that virtualized instances can fail over to another host should something occur with the primary host. Finally, you can migrate virtualized guests from one physical host to another with no downtime, easing servicing, planning, and reorganization, while significantly limiting detrimental effects on production services. You can also take advantage of the new disk quorum features in Windows Server 2008, which allow you to have clusters in multiple locations—say, on both coasts of the U.S., or on different continents across the world—without necessarily having to have a single shared disk between them, something that's required for clustering using Windows Server 2003.
Additionally, you can install Hyper-V on a Server Core installation of Windows Server 2008 and take advantage of the stability and reduced overhead benefits (which are, of course, tangential with high availability objectives) of that style of deployment as well.