[CoLoCo] virtualization in hardy: kvm

Kevin Fries kfries at cctus.com
Mon Mar 17 15:35:07 GMT 2008

On Fri, 2008-03-14 at 23:45 +0100, Soren Hansen wrote:
> Er.. That's hardly surprising. kvm depends on the availability of VT-x
> or SVM extensions available in newer Intel or AMD processors (quite a
> bit of KVM's simplicity can be attributed to this requirement,
> actually). This is *very* well documented.
> What hardware have you attempted this on? Which distribution and
> version? Did you use said distribution's shipped version of kvm?

I am currently using two IBM R40 laptops in the office, but my home
machine is a 2.6GHz P4.  It has HT but not VT.  My computer is ready for
replacement only because I am in the industry.  Most people that I know
that are not in the industry, have older computers.  Since Ubuntu is
such a "end-user" focused distro, is requiring such a modern computer
prudent, especially when other solutions are available that do not have
that requirement?

> > On the few times I fought with it enough to get it running, the
> > performance was better than qemu (and lets face it, that aint hard),
> > but no where near what VBox can do.
> A standard build of KVM will - if the CPU extensions are not available -
> fall back to non-accelerated mode, which *is* qemu. If that feels faster
> than qemu, that's not much more than a statistical anomaly, I think.  I
> have no benchmarks to back this up, but KVM and VirtualBox should
> provide about the same performance.

KVM does not fall back, it fails.  I tried this, and it tells me it will
not run the post install script from the apt-get.

> So your problem with KVM is that it requires hardware support?

Not completely.  My problem with it is that it requires too modern a
computer to run at all, then when it does, it still does not perform as
well as other product that already exist in the same marketplace.

Its almost like justifying Microsoft Windows with the following agument:

"Well sure it cost more, but once you pay for all the software and
hardware upgrades, you will have a system that still is not as stable as

"Well sure you have to upgrade your hardware to a newer rev, but when
you are done, you will have a product that is not a cross platform, not
as fast, and not as stable as the free version of VMWare."

That is my problem with it.  I would rather focus on an industry
standard product that works, and is free, than on that has been plagued
with problems and is open source.  Reality is a cruel mistress.  I also
see this as one of the key problems with Bug #1.  Geeks like me aside,
most people just want stuff that works.  The do not care about open
source.  They see VMWare, VBox, KVM, and Xen all in the same light.  I
do not have a problem with KVM being in the repositories, but VMWare is
the industry leader, and is the solution people are looking for.  That
should be our focus.

Focusing on Linux as a replacement for Windows in VM technologies also
has another benefit.  Microsoft is spending lots of money trying to tell
everyone that Linux is not as easy or cheap to run as Windows.  We all
know that is crap.  But businesses are only hearing one well funded
argument.  Focusing on tools that are cross platform allows the Linux
community to fight the FUD with reality.  By focusing on VMWare, it
gives the Linux community the extremely legit argument of: Go ahead and
try us, its free, and you can use the same VMWare technology you are
used to.  Once they try Linux and find out it is this easy to run, its
easier to say: You know, we make a great border router [mail server,
database server, etc] also.  Its the same embrace and conquer mentality
that Microsoft used against Unix so many years ago... How sweet would it
be to turn the tables.  That can not happen if you are using Linux only

> Er... No. VirtualBox OSE is not the same as VirtualBox. MySQL, however,
> is (not surprisingly) the same as MySQL.

I said nothing about OSE.  Virtualbox is free for persoanal use.  You
may also obtain a version with commercial support... alla MySQL.  VBox
OSE is a completely different project.

You are confusing the goals of Ubuntu with Gobuntu.

> >>> VMWare server is free to use, period.  
> >> If you accept the terms of use and obtain a license key, yes.
> >> Exclamation point![1]
> > So what!!!
> Er... So it's not "free to use, period.". If you don't care about the
> difference between free beer and free speech, that's your "problem".

Again, you are confusing the goals of Ubuntu with Gobuntu.
> > I put this into the same category as Acrobat.  Evince does not do as
> > nice a job, but everyone fawns over it because its open source.  Again,
> > so what!  Instead, I guess I take a more broad look at these types of
> > things...
> I'm struggling to figure out why you work on Linux?

I know the Windows platform very well.  I know the Linux platform very
well.  I use the Linux platform because it is a better solution.  Well
it is except for all the open source is better than commercial crap.  I
take a practical approach.  As such, I have no problems with commercial
packages in Linux.

I believe that oppressive monopolies from Seattle are harming American
business, and leaving the US in situation that  is not a good one.
Europe is starting to get it, and their economy is getting stronger and
stronger.  Meanwhile, the American government refuses to enforce our own
Anti-Trust laws in order to keep the tax money rolling in.  And the
funny part about it, we continue to do more damage to our economy, and
therefore in the long run decrease our tax base, by failing to get real

Real competition comes when companies are all given an equal footing,
and equal chance in the market place.  You can not figure out why I am
in Linux because you are as bad as Microsoft.  In your world, everything
must be better if it is open source, and evil if it is commercial.  I
see a place where both could (and should) live side by side.  Free
software, whether open sourced or not, should always be available to
everyone.  It should be based upon standards that allow all OSes to play
evenly.  Whether that software is open sourced, or a free commercial
product, I have less concerns, as long as it is based upon a standard
that all players must abide by.  Commercial software should allow
feature and usability enhancements upon the free software, or technical
support.  It is only in this way that everyone can be served in a free
(as in speech), clean way.

We start by enforcing standards, real standards like ODT, and not
f(&*^ed up standards like OOXML.  Then we make that standards mandatory.
If Microsoft want to make a copy of MS Word that is free to the public,
and supports ODT as its native format, I say welcome to the party.  Come
on in, the water is fine.  If MS decides that they want to make OOXML as
an optional format so that their users can communicate with their
commercial product, I say fine to that also, just so long as the
standards based functionality is the default behavior, and the OOXML is
only available as a save as option.

Given this philosophy, it should not be hard to see why I am in Linux.
I believe everyone should be given the opportunity to build a better
wheel and put it on the market.  Every product, both FOSS and commercial
should be based upon its merits, and judged by it cost vs benefit.
Personally that is all I am looking at.  You are only looking at whether
somebody will let you look at the code.  I know this is a long running
argument, and Mark has shown repeatedly a practicality similar (though
maybe not as radical) as mine.  The restricted drivers are a perfect
example (See Mark Shuttleworth's blog posts if you want more details, I
will not speak on his behalf).  You are making the Gobuntu argument all
over again.  Its been made, and while not completely rejected, was
mostly rejected... that is why we have the Gobuntu distro now.  Ideals
are great, and the ones you are holding on to I would even consider
noble, but they are not always practical.

> > It is for this reason that I use nVidia video cards, I use Firefox web
> > browser, I use OpenOffice productivity suite, I use Adobe Acrobat to
> > view PDF files, and I use VBox (i.e. to run a Windows instance on my
> > desktop using the seamless mode feature so I can run Visio) and VMWare
> > (generally for virtual servers on a dedicated Linux server).
> What is this seamless mode of which you seem to be so fond?

VBox will allow me to run my Windows desktop on top of my Linux desktop
in one integrated view.  Since I run the AWN launcher, I tell Windows to
"auto hide" its startbar.  When I want to run a Windows program, I run
my mouse cursor to the bottom of the screen, and I get the start button.
Programs->Microsoft Office->Visio, and Visio is running in a window on
my otherwise Linux screen.  All without any complicated network setups.
All I have to do is start VBox (I have it as a shortcut on my AWN bar).
If Windows comes up in the normal bounding box, I just press ctrl-L and
Windows hides itself.

> >>> Of course, all the really cool tools are in the paid version.
> > I don't know about that... 
> Er... You wrote that, didn't you? Or did I get my quoting messed up?
No, I never said that, if I did, it was a mistake.  All the cool
Enterprise tools are in the paid version.  The free version is excellent
for the average user, and Joe Average is who we should be talking about

> > KVM is refusing to load, but VMWare loads, and is running 4 virtual
> > private networks consisting of 11 machines on a single machine with 2
> > cores and 2GB of ram. 
> Good for you. I'm not sure what your point is.

My point is, you are working on a distro that professes its "I just work
out of the box and am so end user friendly" mentality.  Why are you
putting programs in the main distro that are the furthest from that
philosphy.  VBox and VMWare work where KVM does not.  What is so
difficult to understand about this?

> > You are touting a product that will not run, 
> Erm... Please tell that to the 10 virtual machines I have running here
> on my laptop. In kvm, that is.

On your laptop yes, but not on many, many others.  Others that would run
with any other product on the market.  This is the same bogus argument
given by the software developer that installs tons of tools to develop a
program, then distributes it, but was unaware that the tool included a
library it did not document very well (Any other Visual Studio survivors
other than me out there?).  Then they argue, there is no problem because
the program works on their development machine.

Just because it works on your developer machine, or on a few of my
servers (just not the laptops we are taking to Japan), does not mean it
will run on Joe Average's computer.  The reality is, sometimes it will,
sometimes it won't, but VBox and VMWare always will.  So KVM should not
be used and emphasis should be placed on one of these other products.

> I happen to be in the free software business. Your signature suggests
> you are, too? 

I am in the solutions business, and so should you.  Free software is a
tool, not an absolute.  We use whatever provides the biggest bang for
the buck.  Nearly always that is free software, and less often, but
still a majority of the time, it is FOSS software.  But I am _NOT_ in
the free software business... I am in the solutions business.

> It's interesting that you've been able to figure out that whether "the
> vt flag is set" (a rater unfortunate wording, but let's go with it)
> means something for Xen, but you somehow seem to think that whether kvm
> will run is completely random and unrelated to whether "the vt flag is
> set". (Well, not completely random, as you've found that the algorithm
> seems to favour "fail to run" over "ok, go!")

Xen will run if the VT flag is present in the processor... It just will
not run MS Windows in native mode (there are some supposed tricks to
make it work, I just never put that much effort into Xen as I was
unimpressed with it).  KVM on the other hand will fail to load, period,
even if you are running another Linux instance.  Therefore, Xen's
reliance on VT and KVM's reliance on it, are completely different.

> It never ceases to amaze me how you can get completely focused on one
> shortcoming of something and thus conclude that said something is
> completely and utterly broken. For added hilarity, this alleged
> "shortcoming" of kvm is a deliberate design choice of kvm's authors. 

Apples and oranges, see previous comment

Kevin Fries
Senior Linux Engineer
Computer and Communications Technology, Inc
A Division of Japan Communications Inc.

More information about the Ubuntu-us-co mailing list