[Ma-linux] Cloud Computing
chris at servercave.com
Mon Jun 1 15:33:36 EDT 2009
As the two messages on this over the weekend are two different trains of
thought, I'll answer both in kind.
Przemek Klosowski wrote:
> Chris Rogers gave a pretty good narrative describing history and
> future of distributed virtual computing. It'll be interesting to see
> who can deliver best on the promise: VMware has the early lead, but
> Citrix/Xen and Redhat with KVM are on their heels and Microsoft HyperV
> wants to squeeze in too although how will they deal with licensing for
> easy up-easy down virtual machines escapes me. Chris likes Xen, I like
> KVM--we'll see.
Starting disclaimer: I simply despise VMWare. They had a good product
way back when, but refuse to innovate at all with it, are very arrogant
about pricing and partnering, and insist that things only be done their
way. Unless they do some serious reworking of their architecture and
pricing model, look for them to start losing. Could be a slow burn, but
I look for a severe implosion in 1-2 years. Yes, I've had unpleasant
experiences with them, and hence (partially) my biased opinion on the
Outside of that, I think each has their place. I scratched my head for
a long time about Redhat's decision about KVM. I begin to see the
reasoning now. Anyone out there, feel free to correct me. Redhat is of
the thinking that monolithic kernel mods are inefficient and therefore
any modification to the central kernel should go into a module. Hence
KVM. Easy to add, easy to remove. Hmmmm. My biggest thing is
efficiency. Seeing how the Xen paravirtualized kernel can perform and
take full advantage of the hardware virtualization assist extensions on
the chips, that's key for me.
In my sales life, my customers want to know that their apps will perform
just as well as a physical server install. Add the price point to that
(FREE for XenServer, $2500 for Citrix essentials, $5700 for VMWare ESX
VI3) it's easy to get folks started with some great features for free
(and it's easy), and then make a quick license upgrade if they need
things like high-availability, storage centralization, thin provisioning
capabilities, and a lab manager.
The big thing that I want to see is the promised (supposedly)
interoperation with windows/suse/redhat hypervisors and VM's. There
were big announcements followed by dead silence.
> Chris, how do you think Citrix is coping with integrating Xen with the
> rest of their technology/product line? I imagine (and heard rumors)
> that it might be hard because they are quite different in scope and
> target audience. It'd be interesting to see how a proprietary company
> buys an open-source based technology and merges it successfully---so far
> I can only think of IBM succeeding in that,
I had the same initial thought. Citrix? WHAT? Who went on a bender
the night before that? Having been in the mix of it for 18 months now,
I like what I see. They have a CLUE. It's so refreshing to see that
from such a large company. They are integrating the products slowly and
deliberately. They have a good vision, and have had good execution to
date. Serge hits on a point here that I'll hit in my response to him.
> By the way, one enlightening way of looking at the IT revolution is to
> realize that technology evolves from its components being like pets to
> being like cattle. Computers, disk files, network nodes, web servers,
> and whatever else, start as precious, individual entities that we
> remember, give unique names to, etc. Soon enough, however, we acquire
> tons of it, and have to treat as a herd, with numbered eartags rather
> than handpicked names or configurations. Anyone, like Microsoft,
> whose income depends on treating computers as individual entities
> which each have to pay their share of license revenue, will have to
> change or fall behind.
They will fall behind. They will not change.
> Ma-linux mailing list
> Ma-linux at calypso.tux.org
More information about the Ma-linux