VIRTUALIZATION CONCEPT

Yogesh Kumar

AGENDA

Introduction to Virtualization Terminologies of Virtualization Basic Concepts & TypesIntroduction to HypervisorsBare metalKVM as a use case

INTRODUCTION TO VIRTUALIZATIONThe idea behind VMs originates in the concept of virtual memory and time sharing. All of which are concepts that were introduced in the early 60s, and pioneered at the Massachusetts Institute of Technology and the Cambridge Scientific Center.The most popular open-source virtualization suites, Xen, KVM, Libvirt and VirtualBox.

BENEFITS OF VIRTUALIZATION Benefits are as:Server consolidation, hardware cost and performance.Isolation and Ease of Management - Allowing users to have concurrent

operating systems on one computer, have potentially hazardous applications run in a sandbox, all of which can be managed from a single terminal.

Virtualization tools like Qemu and KVM are widely used by Linux developers during their development cycle and testing.

Virtual testbeds for Education.

DISADVANTAGES OF VIRTUALIZATIONDemerits are:Single Point of FailureChances of Performance hitApplication Support

TERMINOLOGIES OF VIRTUALIZATION CONCEPT Virtual Machine (VM) :

A virtual machine is the machine that is being run itself. It is a machine that is ”fooled” into thinking that it is being run on real hardware, when in fact the machine is running its software or operating system on an abstraction layer that sits between the VM and the hardware.

Virtual Machine Monitor/Hypervisor(VMM): The VMM is what sits between the VM and the hardware. There are two types of VMMs. Type-1 and Type-2. Type-1 Native sits directly on top of the hardware. Mostly used in traditional virtualization systems from the 1960s from IBM and the modern virtualization suite Xen. Type-2 Hypervisor sits on top of an existing operating system. The most prominent in modern virtualization systems like KVM, Virtual Box, VMware Workstation etc. The abbreviation VMM can be both virtual machine manager and virtual machine monitor.

EVOLUTION OF VIRTUALIZATION TOWARDS CLOUD

TYPES OF VIRTUALIZATION Type-1 VirtualizationFigure: Type-1 or bare-metal Hypervisor sits directly on host hardware

TYPES OF VIRTUALIZATION CONTINUED..

Type-2 Virtualization:Figure: Type-2 hypervisor runs as an application on host operating system

TYPES OF VIRTUALIZATION CONTINUED.. Operating System level Virtualization:

KERNEL-BASED VIRTUAL MACHINE(KVM) Open Source Solution KVM technology is a kernel device driver for the Linux kernel, which takes full

usage of the hardware extensions to the X86 architecture. KVM allowed for guests to run unmodified, thus making full virtualization of

guests possible on X86 processors. Uses existing Intel VT-x and AMD-V technology, to allow for virtualization: A goal

of KVM was to not reinvent the wheel. The Linux kernel already has among the best hardware support and a plethora of drivers available, in addition to being a fully blown operating system. So the KVM developers decided to take use of the facilities already present in the Linux kernel and let Linux be the hypervisor. KVM virtualization solution for the Linux kernel on the x86 platform.

KERNEL-BASED VIRTUAL MACHINE(KVM) CONTINUED..

KVM developers uses the facilities already present in the Linux kernel(having a plethora of drivers available, in addition to being a fully blown operating system) and Linux acts as the hypervisor.

KVM allows the guests to be scheduled on the host Linux system as a regular process, in fact a KVM guest is simply run as a process, with a thread for each virtual processor core on the guest.

Accepted on the Linux kernel since version 2.6.20. Red hat had “Xen” as foundation for virtualization solution, shifted to KVM with OS version 6.

KVM ARCHITECTURE

KVM DETAILED ARCHITECTURE

All guests has to be initialized from a user-space tool, this usually is a version of Qemu with KVM support.

Each guest processor is run in its own thread that is spawned from the user space tool, which then gets scheduled by the hypervisor.

Each guest process and processor thread gets scheduled as any other user process alongside other processes by the Linux kernel. Each of these threads can be pinned to a specific processor core on a multi-core processors, to allow some manual load balancing.

The memory of a guest is allocated by the user-space tool, which maps the memory from the guests physical memory to the hosts virtual memory.

I/O and storage are handled by the user-space tools

KVM WORKFLOW

QEMU Process Emulator and Virtualizer : Qemu in itself is only a emulator, when

put together with virtualization tools like KVM, it becomes a powerful virtualization tool.

Supports a mix of binary translation and native execution, running directly on hardware.

Access to low-level serial and parallel ports to be able to communicate with the desired hardware.

Thank You