Previous Chapter - Table of Contents - Next Chapter
Virtualization entails creating (in software) an abstracted complete machine environment, under which end user software runs unaware of the physical details of the machine it is running on.
Quite a few approaches to accomplish this mission, but generally involve host operating system and filesystem which directly uses hardware. One or more guest systems run on top (or underneath, depending on how you look at it) of host system at lower level of privilege. Access to hardware such as network adapters usually done through some kind of virtual device, which may access physical device when required.
Many uses for virtual machine. Major ones: more efficient use of hardware, software isolation, security, stability, safe development options, ability to run multiple operating systems at same time without rebooting.
- Understand the concepts behind virtualization and how it came to be widely used.
- Understand the rules of hosts and guests.
- Discuss the difference between emulation and virtualization.
- Distinguish between the different types of hypervisors.
- Be familiar with how Linux distributions use and depend on libvirt.
- Use the qemu hypervisor.
- Install, use, and manage KVM.
In this chapter, will be concerned with creation, deployment, and maintenance of Virtual Machines (VMs).
Virtual Machines: virtualized instance of an entire operating system, may fulfill the role of a server or a desktop/workstation.
Outside world sees the VM as if it were actual physical machine, present somewhere on network. Applications running in VM generally unaware of their non-physical environment.
Other kinds of virtualizations include:
-
Network
Details of actual physical entwork, such as type of hardware, routers, etc. abstracted and need not be known by software running on it and configuring it: Software Defined Networking or SDN.
-
Storage
Multiple network storage devices configured to look like one big storage unit, such as disk: Network Attached Storage or NAS.
-
Application
Application isolated into standalone format, such as container, which will be discussed later.
There are differences between physical and virtual machines that are not always easy to ignore. Eg. performance tuning at low level needs to be done (separately) on both VM and physical machine residing underneath it.
Virtualization has long proud history. Implemented originally on mainframes decades ago:
- Enables better hardware utilization
- Operating systems often progress more quickly than hardware
- It is microcode driven
- Not particularly user-friendly
Later on, virtualization technology migrated to PCs and workstations:
- Initially, it was done using Emulation
- CPUs enhanced to support virtualization led to boost in performance, easier configuration, more flexibility in VM installation and migration
From early mainframes to mini-computers, virtualization has been used for expanding limit, debugging and administration enhancements.
Today, virtualization found everywhere in many forms. Each of the different forms and implementations that are available provide particular advantages.
Host: underlying physical operating system managing one or more virtual machines.
Guest: the VM which is an instance of a complete operating system, running one or more applications. Sometimes, guest also called client. In most cases, guest should not care what host it is running on, can be migrated from one host to another, sometimes while actually running.
In fact, possible to convert physical machines to virtual ones, by copying the entire contents into a VM. Specialized software utilities can make this easier.
Low-level performance tuning on areas such as CPU utilization, networking throughput, memory utilization, often best done on host, as guest may have only simulated quantities not directly useful.
Application tuning done most on guest.
First implementations of virtualization on PC architecture -> through the use of emulators. Application running on current operating system would appear to another OS as specific hardware environment. Emulators generally do not require special hardware to operate.
Qemu -> one such emulator.
Emulator
An Emulator runs completely in software. Hardware constructs replaced by software. Useful for running virtual machines on different architectures, eg. running pretend ARM guest machine on X86 host. Emulation often used for developing operating system for new CPU, even before hardware is available. Performance relatively slow.
Host system (besides functioning normally with respect to software that it runs) also acts as hypervisor that initiates, terminates, manages guests. Also called Virtual Machine Monitor (VMM).
Two basic methods of virtualization:
-
Hardware virtualization (also known as Full Virtualization)
Guest system runs without being aware it is running as virtualized guest, does not require modifications to be run in this fashion.
-
Para-virtualization
Guest system is aware it is running in virtualized environment, has been modified specifically to work with it.
Recent CPUs from Intel and AMD incorporate virtualization extensions to the x86 architecture that allow the hypervisor to run fully virtualized (ie. unmodified) guest operating systems with only minor performance penalties.
The Intel extension (Intel Virtualization Technology), usually abbreviated as VT, IVT, VT-32, or VT-64, also known under the development code name of Vanderpool. Has been available since the spring of 2005.
The AMD extension usually called AMD-V, still sometimes referred to by the development code name of Pacifica.
For detailed explanation and comparison of how these two extensions work, see the Xen and the new processors article.
Can check directly if your CPU supports hardware virtualization extensions by looking at /proc/cpuinfo. If you have an IVT-capable chip, will see vmx in flags field. If you have an AMD-V capable chip, will see svm in same field. May also have to ensure virtualization capability is turned on in your CMOS.
While choice of operating systems tends to be more limited for para-virtualized guests, originally they tended to run more efficiently than fully virtualized guests. Recent advances in virtualization techniques have narrowed or eliminated such advantages, and the wider availability of the hardware support needed for full virtualization has made para-virtualization less advantageous and less popular.
Most modern hardware has hardware virtualization abilities (must be turned on in BIOS).
Hypervisor can be:
- External to the host operating system kernel: VMware
- Internal to the host operating system kernel: KVM
In this course, will concentrate on KVM, as it is all open source, and requires no external third party hypervisor program.
Going past emulation, merging of hypervisor program into specially-designed lightweight kernel was next step in Virtualization deployment.
VMware ESX (and related friends): an example of hypervisor embedded into operating system.
Dedicated Hypervisor
The KVM project added hypervisor capabilities into the Linux kernel.
As discussed, specific CPU chip functions and facilities were required and deployed for this type of virtualization.
Hypervisor in the Kernel
The libvirt project: toolkit to interact with virtualization technologies. Provides management for virtual machines, virtual networks, and storage. Available on all enterprise Linux distributions.
Many application programs interface with libvirt. Some of the most common ones: virt-manager, virt-viewer, virt-install, virsh.
Complete list of currently supported hypervisors can be found on libvirt website:
- QEMU/KVM
- Xen
- Oracle VirtualBox
- VMware ESX
- VMware Workstation/Player
- Microsoft Hyper-V
- IBM PowerVM (phyp)
- OpenVZ
- UML (User Mode Linux)
- LXC (Linux Containers)
- Virtuozzo
- Bhyve (The BSD Hypervisor)
- Test (Used for testing)
Many utilities using libvirt. Exact list will depend on Linux distribution. Full list can be found on libvirt website.
In this course, will work through the use of robust GUI, virt-manager, rather than make much use of command line utilities, which lead to more flexibility, as well as use on non-graphical servers.
QEMU stands for Quick EMUlator. Originally written by Fabrice Bellard in 2002. (Bellard is also known for feats such as holding, at one point, the world record for calculating π, reaching 2.7 trillion digits.)
QEMU: hypervisor that performs hardware emulation, or virtualization. Emulates CPUs by dynamically translating binary instructions between host architecture and emulated architecture.
Host and emulated architectures may be different, or the same. Numerous choices for both host and guest operating systems.
QEMU can also be used to emulate just particular applications, not entire operating systems.
By itself, QEMU much slower than host machine. But, can be used together with KVM (Kernel Virtual Machine) to reach speeds close to those of native host.
Guest operating systems do not require rewriting to run under QEMU. QEMU can save, pause, restore a virtual machine at any time. QEMU is a free software licensed under the GPL.
QEMU has the capability of supporting many architectures, including: IA-32 (i386), x86-64, MIPS, SPARC, ARM, SH4, PowerPC, CRIS, MicroBlaze, etc.
Cross-compilation abilities of QEMU make it extremely useful when doing development for embedded processors.
In fact, QEMU has often been used to develop processors which have either not yet been physically produced, or released to market.
Used by itself, QEMU relatively slow. However, can be integrated with third party hypervisors, and then reach near native speeds. Note: some of these systems are very close cousins of QEMU; others are more remote. A few main ones:
- KVM offers particularly tight integration with QEMU. WHen host and target are same architecture, full acceleration and high speed delivered. KVM native to Linux. Will discuss this is detail.
- Xen, also native to Linux, can run in hardware virtualization mode if architecture offers it, as does x86 and some ARM variants.
- Oracle Virtual Box can use qcow2 formatted images, and has very close relationship with QEMU.
In this course, recommend using (and will use in labs) virt-manager to configure and run virtual machines. Will also give instructions on how to run them using qemu command line utilities.
QEMU supports many formats for disk image files. However, only two used primarily, with the rest being available for historical reasons and for conversion utilities. These two main formats:
-
raw (default)
The simplest and easiest format to export to other non-QEMU emulators. Empty sectors do not take up space.
-
qcow2
COW stands for Copy On Write. Many options. See man qemu-img for more details.
To get list of supported formats:
c7:/tmp> qemu-img --help | grep formats:
Supported formats: vfat vpc vmdk vhdx vdi ssh sheepdog rbd raw host_cdrom host_floppy host_device file \
qed qcow2 qcow parallels nbd iscsi gluster dmg tftp ftps ftp https http cloop bochs blkverify blkdebugNote in particular:
- vdi: Used by Oracle VirtualBox
- vmdk: Used by VMware
Note: qemu-img can be used to translate between formats, eg.:
$ qemu-img convert -O vmdk myvm.qcow2 myvm.vmdkKVM uses the Linux kernel for computing resources including memory management, scheduling, synchronization, and more. When running virtual machine, KVM engages in co-processing relationship with Linux kernel.
In this format, KVM runs the virtual machine monitor within one or more of the CPUs, using VMX or SVM instructions. At the same time, Linux kernel is executing on other CPUs.
The virtual machine monitor runs the guest, which is running at full hardware speed, until it executes an instruction that causes the virtual machine monitor to take over.
At this point, the virtual machine monitor can use any Linux resource to emulate a guest instruction, restart the guest at the last instruction, or do something else.
By loading the KVM modules and starting a guest, turn Linux into hypervisor. The Linux personality is still there, but also have the hardware virtual machine monitor. Can control the Virtual Machine using standard Linux resource and process control tools, such as cgroups, nice, numactl, etc.
KVM first appeared (pre-merge) as part of a Windows Virtual Desktop product. At the time of its upstream merge in 2007, KVM required recent x86-64 processors. On x86-64 platforms, KVM primarily (but not always) a driver for the processor's virtualization subsystem.
KVM appeared as a trio of Linux kernel modules in 2007. When paired with modified version of QEMU (also provided at the same time), created a hypervisor that used the Linux kernel for most of its runtime services.
Shortly after Avi Kivity, the author of KVM, submitted source code to the Linux development community, Linus merged KVM into his Linux tree. This was surprising to a lot of folks.
Many low level commands for creating, converting, manipulating, deploying, maintaining virtual machine images.
Managing KVM can be done with both command line and graphical interfaces.
Command line tools include: virt-* and qemu-*. Graphical interfaces include virt-manager, kimchi, OpenStack, oVirt, etc.
As you develop more expertise, will become practiced in using them. But, for all basic operations, virt-manager will suffice and that is what we will use.
