During our studies for the recently minted Cloud Computing Infrastructure subject at UTS, we studied the Kusnetzky Group Virtualisation Model. Developed (surprising though it may seem) by the Kusnetzky Group, it identifies several distinct layers of offerings beyond those of the somewhat overused NIST Service as a Service acronyms.
A terribly brief rundown
Most of the layers seemed pretty self explanatory:
Application virtualisaion, where software is developed to be deployed on a common application framework. Through a bit of a logical stretch, I suppose the JVM could be included here too.
Processing virtualisation, where hardware and software are abstracted, divvied up and served independent of their physical implementation.
Storage virtualisation, where file systems, physical hardware and their interconnects are implemented to serve a pool of logical space. In the NIST definition, this and the preceding layer would be combined to form IaaS.
Network virtualisation, where networking hardware and software are abstracted. More specifically, where logical networks are independent of their physical implementation.
These actually do run down, though
Intersecting these layers like syrup dripping and running down a lovely stack of soft pancakes, we have the management layers:
Security for logical environment, where policies and technology serve to secure (surpsiring though it may seem) the virtualised infrastructure.
Management of logical environments, where the deployment of the virtualised services is automated and allowed to be served as a single pool of computing resources.
Ruben, you missed out access virtualisation
Which brings me to the first and highest layer in the Kusnetzky Virtualisation Model. The other layers seemed to fit into place in my brain fairly easily. I did like how the management layers were articulated too, though I think a specific provisioning layer separate from management would make sense.
What had me scratching my head was access virtualisation. I assumed it had little to do with virtualising a Microsoft consumer database. Oh Ruben, you so funny.
From what I've been able to research, access virtualisation has to do with the hosting of software on a server, and the rendering of said software on a separate client device. This client device is able to interface with the application, and determine how best to render it. In this way, it's similar to the NIST definition of SaaS, but with perhaps a little more logic on the client side.
There could be circumstances where that distinction is important. In the case of phones, a user could choose to view a phone, tablet or desktop optimised version of a web application, or even use an installed thick client application that calls the server hosted software.
I could be misunderstanding this distinction; I'm certainly open to correction. If it turns out there is no such distinction, perhaps I should create a Ruben Sandwich of Virtualisation model. That has a nice Burger Ring to it.