Paul,

Thanks for your comments. You are, of course, correct that virtualization started with the mainframe. I am lucky enough - or old enough? :) - to have worked on mainframes and have blogged on the very same thing in the past (http://blogs.egenera.com/pete_manca/).

My point in this blog was that server virtualization is only part of the story in this new emerging x86 virtualization market and other complementary technologies, like IO virtualization, are just as important, if not more important, when trying to create a dynamic data center. In reality, the mainframe has been doing server virtualization and IO virtualization for years. It is just not that relevant in the discussion today around the emerging trends in data centers. That is not meant to be negative on the mainframe, a platform I have immense respect for, however, it is not really in the mix when discussing new wave virtualization technologies.

Pete

To expand on a feedback I posted on 9 January to another similar article:

IBM commercially introduced virtualization in 1967, it's not the "next" big thing, it's been around for over 40 years. It JUST recently (even "recently" is a
relative statement over 40 years) was embraced and exploited by the x86 world, both AMD and Intel are finally stepping up and putting the hardware features in to support it better, features that have existed on other platforms for years.

What's more interesting today is to realize is that there are three logical virtualization layers:

1. HW virtualization (goes back to 1967) includes not just the processor and memory, but also storage, network, I/O, and more.

2. Middleware virtualization (goes back to the early 80s at least, unless you want to count IBM's two premier transaction managers CICS and IMS which are both 40 or more years old and provided for virtualized transactional services long before the 80's)

3. Service (SOA) virtualization, more recently formalized for Enterprise specific services.

Net is virtualization is finally "reaching" platforms like x86, but it is still no where near as sophisticated (and capable) as it has been for, in some cases, decades on platforms like IBM's System z and even IBM System p. What I find interesting is to watch these more mature systems and learn what the distributed world has to "work up to" someday.