Target Windows?

Posted Jan 26, 2007 9:57 UTC (Fri) by AnswerGuy (guest, #1256)
In reply to: An introduction to lguest by sbergman27
Parent article: An introduction to lguest

... Unless you're using target in a more militant sense then I don't see how the statement logically follows?

Linux has too many virtualization models. Factions of developers are experimenting and filling different virtualization niches.

VMware is commercial, expensive, proprietary, and well established. Full virtualization via some proprietary (and patented?) software techniques. Adding support for para virtualization. Fairly slick UI and provisioning tools.

Xen is free. But separate (and widely criticized). Supports para virtualized and full virtualization (with the latest CPU models). Nascent tools

KVM is free ... and done by core kernel developer. Thus it's integrated into the mainstream (but perhaps too new to have accumulated much criticism). No tools nor UI t speak of --- KVM is more of an API and some other project(s) will have to form around that to build front end tools and UIs. KVM seems to require the VT-x CPU features in any event.

Virtuozzo and OpenVZ are very lightweight but essentially like a jazzed up "chroot" (or, more to the point, like the FreeBSD jail() system call). It's well established for "consolidated server hosting."

This lguest, as described here seems to be an experiment in minimalism.
Interestingly it using the ancient 386 segmentation and additional "rings"
as the basis for it's own paravirtualization abstraction. I'm guessing
it will be somewhere between Virtuozzo and KVM in the degree of
isolation and virtualization that's achieved.

So all of these other than VMWare sound like perfectly normal exploration of a range of possibilities. I suspect that parts of these will be
merged ... but that there will continue to be at least two or three diverse set of "virtualization" and "jail/chroot isolation" features that will
be used in different ways.

More importantly I hope that provisioning, package management, monitoring and related tools become "VM aware." in a large number of ways. And that they can mostly abstract away the differences between Virtuozzo, Xen and other VM types.

JimD

And not forgetting UML

Posted Jan 26, 2007 11:12 UTC (Fri) by copsewood (subscriber, #199) [Link]

Which is User-Mode Linux. On which I've run an incredibly reliable and low cost server doing a great number of things reliably for a couple of years. UML has been criticised for performance compared to Xen - but most servers only use a small fraction of CPU anyway. Don't notice the problem myself. Virtualisation has so much potential it's not surprising there is much work going on different approaches to it. On Linux it is also a young enough field we are all better served by finding out which solutions work best in which situations.

Target Windows?

Posted Jul 30, 2007 5:12 UTC (Mon) by micah (guest, #20908) [Link]

Why do Linux-Vservers always get left out of the list of virtualization technologies? Even moreso, why does Virtuozzo and OpenVZ get mentioned but not Linux-Vserver?

(追記) (追記ここまで)