🐏 mogoz

Search

SearchSearch

Virtualization

Sep 16, 2024, 4 min read

tags : Operating Systems, Orchestrators and Scheduling, QEMU

Intro

  • Virtualization: creating and running virtual machines
  • It is handled by something called a hypervisor
  • Hypervisor can either be:
    • software
    • firmware
    • hardware

Types

Type 1

AKA True virtualization

  • Run on bare metal and typically leverage features of the CPU specifically built for virtualization. Eg. AMD-V and Intel VT-x.
  • Can only emulate the same architecture as of the host CPU
  • Examples
    • KVM/Proxmox (Not exactly)
    • VMWare ESXi.
    • Microsoft Hyper-V

Type 2

AKA Emulation

  • Run on top of a host OS, and thus it translates system calls made by the guest OS to system calls made to the host OS.
  • Can emulate any and all architectures regardless of the host CPU, at least in theory.
  • Examples
    • QEMU. (Not exactly) See QEMU
    • VMWare Workstation.
    • VirtualBox.

Case about QEMU+KVM

  • QEMU uses KVM
  • QEMU and KVM are neither type 1 nor type 2; somewhere in between.
  • If KVM Module
    • Present: QEMU will leverage KVM and act halfway between type1&2.
    • Absent: QEMU will provide emulation rather than virtualisation.
  • About KVM itself, it’s not strictly a type-1 hypervisor, as it still is part of an OS,

Ecosystem

Libvirt

Libvirt is an open-source set of api, daemon, cli and libraries which provide a single way to manage multiple different hypervisors It can handle QEMU, KVM, Xen, LXC, OpenVZ, VMWare ESXi, etc.

Some Notes

  • VMWare offers graphics acceleration for Windows guests that seems to be unmatched in KVM (without using PCIe passthrough).
  • VirtualBox has a basic 3D acceleration support for Windows guests that KVM/QEMU still doesn’t have without GPU passthrough.

Others

CPU Pinning

It moves processes from the host away from certain CPU cores, while devoting the guest entirely to these cores. This way, the ‘pinned’ CPU cores are dedicated to the VM,

Virtual Machines (The ISA one)

In fact, x86 seemed to be doomed at the turn of the century. Much was made of perceived performance boundaries of the x86 ISA; the most common prediction being a shift to “very large instruction word” (“VLIW”) architectures. There was a scramble to find a way to build a compatibility bridge from x86 to this VLIW future.

The 1997 DAISY paper, 1998’s “Achieving high performance via co-designed virtual machines”, and 2003’s “LLVA: A Low-level Virtual Instruction Set Architecture” originated the idea of a “virtual instruction set architecture” (“virtual ISA” or “V-ISA”) to address the inflexibility and seemingly imminent obsolescence of x86. They were written in reaction to Java’s virtual machine; the LLVA paper in particular defines several useful design goals for both the “Virtual Instruction Set Computer” (“VISC"") and “Virtual Abstract Binary Interface” (“V-ABI”) in order to differentiate their approach from the JVM.

Isolation and Sandboxing

Remote code execution/ Sandboxing (Untrusted)

At the moment, this can be one of 2 ways largely

WebAssembly

See WebAssembly

VM

Resources

Resource

TODO Image Types

ISO

.img

qcow

Graph View

Backlinks