you clarified a bunch of thing for me and that just awesome

cheers

With KVM you nearly get all cool linux features like powersaving for example ( Green-IT … buzzword bingo ).
It is nice to see that easy, clean, fast, well supported solutions can win ( sometimes .. SELinux seems to be winning too ).

It’s certainly simpler to stand back, think that all companies that are in business to make money are evil, and play the “I might as well not even try” card. But until you actually do what you’re talking about and produce the same results that you’re prognosticating, you’re just generating FUD. I understand your pessimism, and I have certainly seen my fair share of corporate stupidity, but I have not yet seen anyone actually do a thorough performance test according to VMware’s guidelines, publish them, and then get told to take them down because the results weren’t favorable for VMware.

VMware reserves the right to refuse publication of any benchmark

So ya, I can publish benchmarks, if VMWare is OK with it. Which can mean practically anything. If the benchmarks are unfavorable towards ESX VMWare could simply claim the the methodology isn’t good and therefore the results won’t be published, or just give no reason at all.

So internal testing could be done, but going beyond that to verify methodology (outside of VMWare verifying methodology) simply can’t be done unless the results happen to be favorable for VMWare. And thinking that VMWare will knowingly allow publication of results that make it look worse than the competition would be very naive.

I looked at ESX a bit more closely and I think the main value lies in the software ecosystem around it. It is like Red Hats virtmanager and then some .. but RHEL 6 isn`t that far down the road so I am not really interested in trying a closed source solution with an uncertain future ( tbh I feel the same about Xen .. KVM is so much better NOW, it just needs more support tools ).

Despite my misgivings about the future of virtualisation and where it’s heading with regards to software like VMware and the whole ‘the hypervisor is the OS’ stuff, which I think is nonsense really, being able to pick up ESXi and then move easily to ESX when they’ve tested the water will enable VMware to pick up a lot of customers. I might still be one of them. In terms of well tested ‘enterprise class’ (hate that phrase) stuff then VMware and ESX is still the best around there is at the moment. KVM has matured spectacularly in a short space of time, and I do think they have the right approach with hardware virtualisation happening at the right time (leaving I/O virtualisation), but it really depends on what management tools emerge around it. I’m not enamoured with Red Hat’s. It will be some time yet.