The Virtualization Practice

Posted By Texiwill on December 4, 2009

With the announcement of V-Block and Cisco UCS as a major component, is more hypervisor functionality going to end up in hardware? UCS adds some interesting features into the hardware that were traditionally within the purview of the hypervisor. Now it looks like V-Block is the assembly of myriad components that taken as a whole look remarkably like the beginnings of a hardware based hypervisor.Hypervisors are hybrid network, storage, and compute resources. So in essence V-Block adds ready-made hardware extensions to these core hypervisor functionality.

• Compute -> UCS Blades w/Intel-VT
• Network -> UCS Virtual NICs on the midplane, plus the Nexus switching
• Storage -> EMC Storage with PowerPath Multi-Path-Plugin

More and more hypervisor functionality is being placed into CPUs and hardware. At the moment, there are really only two components that appear to have hardware analogs, but they are related to the third. These two items are the CPU (Compute), and Network. Yet, Storage also uses the network with FCoE so we can start to see the third have hardware analogs soon.

But you say, this hardware always existed so what in the world am I talking about?

Actually, the Intel CPUs have been steadily adding more and more functionality to allow hypervisors to push more and more of its functionality into the CPU, where it runs faster. Intel-VT moved some very complex code that was consistently run within Hypervisors down into the CPU, this trend will not stop. The question is which hypervisor will appear in CPU; Will it be Hyper-V, VMware, or Xen or will Intel and AMD stick to providing tools for all hypervisors equally?

However, with the advent of UCS, Cisco has once more taken control of the virtual switch by introducing two things, the Nexus 1000V and the UCS midplane virtual NIC in hardware. These two logically place networking once more in the hands of the network engineers and not the virtualization administrator. The UCS midplane virtual NIC hardware allows any blade within the UCS chassis any number of virtual NICs for its use.

VMware, EMC, and Cisco have a head start on moving this functionality to hardware, however, this will still take some time, there are a few hurdles still to cross.

Related posts:

1. Virtual Thoughts: Is the Hypervisor moving into Hardware?
2. VMware vSphere Security Guidance, just the Beginning
3. Hosted Virtualization Security – Type 2 Hypervisors
4. EMC/Cisco/VMware Vblock Reference Architecture – Some Storage Details Finally Emerge for Vblock 1
5. EMC/Cisco/VMware vBlock – The Data Center Virtualization Perspective

Categories:  Data Center Virtualization
Tags: Hardware Hypervisor, UCS, v-block