• Table Of Contents

Introduction

Basics

<aside> 👉 What is Virtualisation?

</aside>

• It provides abstraction on the top of actual resources we want to virtualise.
• Based on the abstraction, there is VM based and container based virtualisation. There are other techniques too, such as Unikernels, which are lightweight single purpose VMs

<aside> 👉 VM-based Virtualisation

</aside>

• This approach virtualises the complete hardware, and the abstraction it presents to the VM are virtual disks, virtual CPUs, virtual NICs etc. OS on the VM can continue to do I/O operations on disk, thinking it’s the only OS running on the top of it, but it’s not, the Hypervisor is managing their access to the hardware.
• In a traditional computing model, a single piece of hardware is only supposed to run only a single operating system. The idea behind virtualisation is being able to run more than one OS'es on the same piece of hardware concurrently, all of which are sharing physical resources, and stuff, at the same time.
• It has a lot of benefits such as maximising hardware utilisation, decreasing hardware costs, reducing power consumption, requirement of less workforce, simplification of system management and security.

<aside> 👉 History of Virtualisation?

</aside>

• It all started with an IBM engineer started working on virtualizing the IBM mainframe CP-40, which then evolved into CP-67 which made use of partition technology to run multiple applications at once.
• Then came the UNIX which supported multi-programming on x86.
• Then came Sun Microsystems’s Java, which brought the write-once-run-anywhere notion, with the help of JIT, bytecode and JRE. This was the advent of process-level-virtualisation, where the JRE virtualised the POSIX layer.
• Then VMware stepped into the game and virtualised the actual hardware instead (memory, disks and so on), and they launched ESX (Type 1) and GSX (Type 2) hypervisors.

<aside> 👉 Container-based virtualization

</aside>

This technique doesn’t abstract the hardware but uses techniques within the kernel to isolate access for different resources viz. different network subsystem, different process tree, separate file system etc.

Host and Guest OS

<aside> 👉 Host Operating System

</aside>

• The host OS boots up the hardware and make the resources available to others for virtualisation
• It can be either a Baremetal Hypervisor (VMware ESXI) or just a normal OS (Windows 10, with VMware Workstation installed)

<aside> 👉 Guest Operating System

</aside>

• The OS'es running in the virtualized environment are known as Guest OS
• They have their own memory allocation, virtualized network adopters, storage devices and other hardware components, all of which is emulated by the Hypervisor
• An OS expects full control over the hardware but the VMM must multiplex hardware access b/ many multiple guests, it can be achieved by tricking the guest into believing it has full control, but in reality it does not.

Hypervisor

Introduction

Basics

A special piece of software used to virtualize the OS, and it consists of two parts:

• Virtual Machine Monitor (VMM): Used for trapping and emulating the privileged instructions set viz. it controls the hardware-software interface.
• Device Model: Used for abstract and emulate the I/O devices, by trapping and emulating them, and delivering interrupts back to the VM.

What does a VMM do?

<aside> 👉 Running multiple OS’es on the same piece of hardware

</aside>

A VMM multiplexes multiple machines over the same hardware, just like how context switching is done by the OS, but now instead of processes, it’s the guest machines which are getting switched from time to time.

But an OS expects its Kernel to run in Ring 0 and have unrestricted access to the hardware, but that’s not possible in the case of guest machines, because we can’t allow a guest to do privileged operations because a malicious one might take over the hardware.

<aside> 👉 Isolation

</aside>

The VMM must isolate guest OS’es from one another

The VMM must make sure the performance is as native as possible, by running majority of the instructions directly on the hardware and only trapping privileged instructions.

<aside> 👉

Ensuring performance

</aside>

<aside> 👉 Handling interrupts and exceptions

</aside>

• The Hypervisor must not allow the guests to execute any privileged instructions, rather trap and emulate them.
• Example: if a network call is issued within the VM, the VMM must trap it, On receipt of a response over the physical network/ NIC, the CPU will generate an interrupt and deliver it to the actual virtual machine that it’s addressed to.

Categories of Hypervisors

<aside> 👉 Baremetal Hypervisors (Type 1)

</aside>

• Directly installed on the physical hardware and runs on top of it to run like an OS itself, to perform virtualisation.
• More popular and secure than its counterpart
• Has lower latency
• Example: Citrix XenServer, VMware ESXi, Microsoft Hyper-V

<aside> 👉 Hosted Hypervisors (Type 2)

</aside>

• They run internally on the top of an operating systems
• Rather than being run as the OS for hardware emulation, they are installed as applications on a Host OS.
• Example: VMware Workstation and VirtualBox

Type 1 and Type 2 Hypervisors

VMM design techniques (CPU virtualisation)

Trap and Emulate

<aside> 👉 Basics

</aside>