4: ESX: Hypervisor-based virtualization

VMWare Inc has a long experience record for x86 (PC) based virtualization.They began with the software/userspace paradigma (see part 2 of these texts). Then they presented the first ESX server - using a Hypervisor paradigma.

Here's a typical ESX system:

virtualization-esx-60.png

The biggest difference to the Xen System (see part 3): There are no paravirtualized domains - every virtual machine has a virtual BIOS and an emulated PC infrastructure. Even the service console runs a slightly modified normal Linux kernel. The service console does the management of the hypervisor (ESX kernel) and collects statistics as well as monitoring the virtual machines. It does however NOT do any storage or network work like in the Xen Dom0 case. All emulated hardware for the virtual machines is emulated by the ESX kernel - achieving much better performance than the "qemu-dm"-processes.

As I said there are no paravirtualized machines, but there are paravirtualized devices! What's the difference? These "paravirtual" devices show up on a normal emulated PCI bus in the guest systems but they offer a more direct connection to the ESX kernel reducing significantly overhead. From the beginning there was the vmxnet paravirtual network device (with its incarnations vmxnet2 and vmxnet3 (since ESX 4)) and (since ESX 4) the paravirtual SCSI controller (PVSCSI). Drivers for the guest operating systems are available with the "vmware tools".

Access to the ESX host is controlled by the vSphere console (click to enlarge):

vmware-esx-demo.png

In this screenshot you see console access to a virtual Windows 2003 machine and a virtual Linux machine.

Like in the Xen case these graphical consoles are not for work but only for access to management functions of the guest operating system not accessible remotely (critical system updates et.al.).

With vmware ESX you get a bunch of management tools and migration support. A feature called "Vmotion" allows you to move a virtual machine seamlessly to another ESX host without interruption of service - with some limitations sure (like identical CPU ids and configuration of the two involved ESX hosts).


  1. Why virtualization?
  2. Userspace-based virtualization (the easy way)
  3. Xen: Hypervisor-based virtualization
  4. ESX: Hypervisor-based virtualization
  5. Solaris Zones: A sharing approach
  6. Hybrid methods: KVM


0 TrackBacks

Listed below are links to blogs that reference this entry: 4: ESX: Hypervisor-based virtualization.

TrackBack URL for this entry: http://southbrain.com/mt/mt-tb.cgi/99

December 2015

Sun Mon Tue Wed Thu Fri Sat
    1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30 31    

About

This blog is owned by:

Pascal Gienger
J├Ągerstrasse 77
8406 Winterthur
Switzerland


Google+: Profile
YouTube Channel: pascalgienger