Talk:Hardware-assisted virtualization

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Wikipedia being used for marketing?[edit]

I have some concerns about this article as I feel is it being used as a marketing mechanism for VirtualIron. A quick google of "native virtualization" reveals no reference to what this article is talking about. In fact, the first time I saw this term is on VirtualIron's blog [1].

I know of absolutely no literature that establishes this terminology. VirtualIron also, not surprisingly, shows up in google adwords when searching this term. Anyone know what the proper way to deal with something like this? --Anthony Liguori 20:40, 7 August 2006 (UTC)

I decided to rewrite this page. Instead of treating "native virtualization" as a new concept (which it isn't), I tried to relate it to "hybrid virtualization" which more established terminology for this approach. I've also attempted to list both the pros and cons of it to meet the NPOV standard (the previous article only spoke of benefits). Feedback is appreciated.--Anthony Liguori 20:56, 7 August 2006 (UTC)

This whole page makes little sense to me. It's not clear that "native virtualization" describes a distinct or distinguishable technology. It appears to describe a hypervisor architecture with some hardware acceleration, an approach that all virtualization vendors appear to be moving towards X86 virtualization. How is this different than something like VmWare ESX (which doesn't "paravirtualize" leveraging Vanderpool or Pacifica processor enhancements? I'd say this is a marketing term, and one that isn't clearly defined. Also, this statement appears to violate NPOV: "In x86 virtualization, the primary implementor of native virtualization is Virtual Iron[1]." What makes them the "primary"? Does this mean that they are the only implementor, or the first implementor, etc? Also, I'm not sure "hybrid virtualization" any more established? There is no wikipedia entry on it, and a quick google didn't return any good matches. --Overunder 15:19, 6 November 2006 (UTC)

The real problem I see here is that this and several other pages linked by virtualization are really just subsections of that page. They are dealing with subtopics of that particular treatment of the concepts. This is the way one breaks up a web page, but not (I think) how to structure Wikipedia entries.
They also all seem very much geared toward today's x86 virtualization issues (and why not? that's the focus of most people's interest) but virtualization has a long and important computer science history. See CP-40 where I have tried to start turning over the rocks on this topic. Full virtualization of the S/360 was a milestone, and that work created many of the concepts we rely on today. The work done in the 60s and 70s may seem like a footnote to what's of interest today, but note that some of the same people have been continuously involved in the technology (see Talk:Hypervisor for instance).
Anyway, I think that these sub-pages perhaps deserve the deep six. I applaud breaking up long pages, but this needs to be done based on universally-recognized concepts. I think the attempt at taxonomy is interesting, and probably a valid way to break up the concepts; but you can't just invent neologisms and put them in the index. At best, you define new terms the way you name variables and lemmas – in the context of a cohesive argument or thesis. Trevor Hanson 05:41, 19 November 2006 (UTC)

I rewrote most of this article. Adding better descriptionn to what Native virtualization. Explaining the history behind full virtualization and paravirtualization.

Proposed opening paragraph[edit]

Here is a proposal for a new opening paragraph, intended to address the context problem.

Native virtualization is a type of virtualization where the virtual machine (VM) simulates the complete hardware, permitting the operation of an umodified operating system for the same type of CPU to execute within the virtual machine container in complete isolation.

Native virtualization refers to sytems and CPU's that meet the requirements for virtualization natively, without any additional support. For example, the 386 family does not, because the instruction to read the current privilege level is not a trapped instruction, resulting in a native VM failing to hide the fact that it is no longer the real hardware.

Keybounce 03:30, 23 January 2007 (UTC)


Comment[edit]

If this native virtualization sections/virtualizes hardware with x86*, why doesn't someone go further and allow for two (or more) OS's to run "natively" on a multiprocessor system (or smp) such that each OS has their own respective cpu, memory, and device sets, allowing for hotswapping of actual hardware with virtual dummy soft-hardware (ie: 1 OS access display card at a time). —Preceding unsigned comment added by 75.178.178.53 (talk) 10:13, 7 November 2007 (UTC)

Indeed, that's possible today. And each guest OS can have several CPUs, not just one. --Gribeco (talk) 19:29, 27 June 2008 (UTC)

Re: Some clarification on commercial aspects and marketing[edit]

There is a strong marketing element in this article.

Large corporations attempt to dupe to manipulate the unwary into a locked-in silos of expense and incompatibility. Do not be conned! Open Source is better if you are in doubt.