Sweet gimme a bit and I'll be throwing some code commits your way
R
On 1 Oct 2010, at 17:42, Richard Gass wrote:
On Thu, Sep 30, 2010 at 1:09 PM, Robert Gibbon
<[email protected]> wrote:
Hi team Tashi
I just found your project and it looks really interesting, albeit
in the
early phases. I've got some questions for you, if you've got time,
maybe you
can answer them...
* I can see that all of the nodes need to have access to a common
backend
filesystem to get the machine images. How difficult would it be to
leverage
a pxe boot server instead? What problems can you see with that?
Yes, there needs to be some system that holds all the images.
* What are your thoughts on dynamic bare metal provisioning (hence
the above
question)?
We use a system that we call Zoni to do bare metal provisioning of
systems. Zoni handles provisioning, management, allocation,
isolation, and debugging, It is currently checked into the tashi tree
(look in the src dir). I am currently putting the finishing touches
on the first working release.
At my work we've got lots of (old school) vertically scaling
implementations
that simply don't perform to the required specification under a
hypervisor
(we've proven it). We can't be the only organisation with that
problem.
I remember chatting to a cloud services vendor who said outright that
running a busy rdbms under a hypervisor is a very very bad idea.
My humble, bare metal cloud provisioning in addition to virtual
provisioning
is definitely the way to go.
Last question, are you looking for some help?
We are always looking for help by way of code contributions. Please
send your commits my way :D
Richard
Regards all
Robert
--
Richard Gass
