http://linux.slashdot.org/article.pl?sid=07/08/14/1618241&from=rss

http://www.venturecake.com/the-vmware-house-of-cards/

Funny, vmkernel does not need linux kernel to run at all! Even in
current 3.x version of ESX server, the console OS is a special VM
running under vmkernel. Do some study.

Some excerpts to debunk that:
(taken from slashdot comments) I think we will hear a bit more about
this.....

by laray88 (903645) on Tuesday August 14, @03:23PM (#20228677)
yes- this is correct. the redhat part kicks off the vmkernel. which is
distinctly written by VMWare. may be based on linux, probably is. but
unless you have the source code , who knows (except vmware)

Re:directly on the harware = uses GNU/Linux
(Score:2)
by rubycodez (864176) on Tuesday August 14, @04:43PM (#20229765)
vmware sales and marketing say "it's just the service console". but
the truth is vmware kernel is a process under control of the linux
kernel. yes, the virtual machines run under the vmware kernel, but the
vmware kernel must use Linux devices and linux virtual memory system
for resource usage of the real hardware. in other words, linux is in
charge, "just a service console" my ass.

Red hat has the right to access vmware ESX server's source code. I
didn't see any complains from Red Hat. In ESX 3 the vmkernel loads
before linux kernel and that is a fact. Now you tell me the Linux
kernel controls the VMkernel? How?

The VMkernel is depending on Linux as a bootstrap, also the Service
Console IS running as a VM.
The Service Console is only running the protocol daemons to have
communication with the outside world. These daemons manage the
settings in the VMkernel. Consider the Service Console as a VM with
additional permissions to the VMkernel. Hypothetically speaking if the
VMkernel was running inside the Linux kernel, how would you explain
the fact that you can change the memory reservations and CPU resource
priority of the Service Console from within the VMkernel settings? You
actually can kill the Service Console VM by setting ridiculous
parameters. Eventually in this situation the VMkernel will die,
because the monitoring by the Service Console is still very important
for a good operation of the VMkernel.

Another argument; Why would anyone think that the VMkernel is Linux?
The scheduler is no use for context switching Virtual Machines. Even
Xen and KVM implemented different schedulers for running VMs.

The VMkernel is only 34MB total in size and is mostly written in
assembly. The VMkernel is using modified Linux drivers supplied by
Redhat and hardware vendors. All drivers have been approved/licenced
by the vendors that are on VMware's HCL.
All changes made by VMware to the Linux kernel and open source
products are published again as is required by the licenses of open
source products used by VMware. Try searching for 'vmware' on
freshmeat.net.

This whole discussion seems like a load of FUD to me.

Exactly, this sounds like FUD shots off the bow from the likes of
Microsoft. I know they have a similar technology coming ("free") with
their next-generation servers.

I think the Microsoft boat has too many leaks to ever be patched up
and will slowly watch their server hardware die a long and terrible
death. Every server administrator I know now virtually refuses to
implement Windows servers in any enterprise solutions.
Having a powerful contendor like VMWare in the Linux VM market is
exciting. They may have trouble competing long-term with Zen. But, I
think VMWare's history, mass-adoption (thanks to the free server
product), and ease of use will let them (and US!) ride this one
through and profit greatly.