concept KVM in category openstack
This is an excerpt from Manning's book OpenStack in Action.
Throughout this book, the Kernel-based Virtual Machine (KVM) will be used. KVM has been part of the Linux kernel since the 2.6.20 release in early 2007, and it’s fully supported by OpenStack. KVM also provides paravirtualization, which must be either supported natively by the operating system or added through the use of hypervisor-specific drivers installed on the virtualized operating system image. The traditional problem with open source hypervisors is that there’s a steep learning curve for deploying and maintaining them, often requiring experience in system, network, and application administration. In organizations lucky enough to have centrally supported virtualized resources, the resource request process must pass through the organization’s network, systems, security, and financial elements in the provisioning process. This typically leaves users with three choices:
KVM (Kernel-based Virtual Machine) —KVM itself doesn’t perform emulation of hardware. KVM is a Linux kernel module that interfaces directly with processor-specific virtualization extensions to expose a standard \dev\kvm device. This device is used by a host monitor, like QEMU, for hardware offloading of emulation functions. What is KVM?
When you hear someone say they’re running KVM, what’s really happening is that they’re using KVM for virtualization-specific hardware offloading and leveraging QEMU for device emulation. When KVM extensions aren’t available, QEMU will fall back to software emulation, which, although it’s much slower, will work.
Unless specified, I’ll refer to the Libvirt, QEMU, and KVM suite of software collectively as KVM.