On 11/20/2012 10:50 PM, Matt Wagner wrote:
On Tue, Nov 20, 2012 at 04:54:51PM +0100, Tomas Sedovic wrote:
> Good news everyone!
>
> During our discussions about the Aeolus API at the first Technical
> Cabal meeting, Michal Fojtik suggested we look at the CIMI spec
> (which Deltacloud partially implements).
>
> Apparently, it tackles a lot of things we are or will be discussing
> here, too. Things such as state changes, multi-instance deployments,
> networking, etc.
This looks interesting. I'd had only brief exposure to the concept
prior to now, but some of the concepts it exposes do seem to line up
well with what we do.
Is Deltacloud still the only project implementing this in the wild? It
seems interesting, but I'd hate for us to implement an API that few
people end up using. (On the other hand, I suppose everyone else is
probably saying this exact same thing right now...)
I don't necessarily suggest we adopt CIMI as-is (though that may in fact
be something we should consider). It's just that it may give us a way to
resolve some of our API design questions.
> You can download it from the DMTF website:
>
>
http://dmtf.org/sites/default/files/standards/documents/DSP0263_1.0.1.pdf
>
> It's a *massive* document (178 pages, ffs) but I think it may be
> worth checking it out.
Wait, you open with "Good news everyone!" and close with a link to a
178-page API specification? I'd hate to see your idea of *bad* news. ;)
To quote Professor Farnsworth from Futurama: "Great news, everyone!
You'll be delivering a package to Chapek 9, a world where humans are
killed on sight."
Seriously, though, this is interesting and seems like it's worth
exploring further. (I can't say I've read the document in full yet.)
-- Matt