Re: The Business of Building Clouds

Hi James,

On 4-Jun-08, at 10:15 PM, James Urquhart wrote:

  Google AppEngine is intentionally a completely different animal than Amazon EC2/S3.  So "pointing the deploy gun" makes no sense whatsoever *unless* you build a AppEngine clone on EC2/S3.  But then you are just building a new stack.  Do we say everyone must develop to the lowest common denominator? 

 I'd turn it around and suggest that perhaps its greatest strength is also its greatest problem.  It's a "Big Monolithic Black Box".    Now, in my experience, black boxes are quite valuable, in they focus you on "getting things done" at a usable level of granularity.   But sometimes you need to be able to see what's behind the curtain to diagnose problems, or to change the way things work a bit.   Hopefully it's well layered or abstracted enough that you don't have to dig into a rat's nest to get anything done.     

It could be that Google wants to be in the "full stack support outsourcing business", but given how laborious that can be, I'd bet they would want to automate a lot of that and turn many of the options into self-service controls.   

Look, that kind of compatibility between any two vendors serving the same layer of David's model makes some sense to me.  However, what we have today is a primordial ooze of technologies, target markets and business models.  I think we need a lot more time before the dominant complex life forms evolve, and the kind of portability mentioned here is "boil the ocean" complex. 

Well, not really, given there's already an OVF draft spec out there that covers a big chunk of what would be useful, though it's virtualization-centric.   

 I would expect every instance-level cloud provisioner out there will have some kind of spec or API published now or in short order, leading to lots of silo's and a need for interoperability.

In short, sometimes one has to be naive and try to drive agreement where the domain is well-understood, like provisioning *nix instances.    This does, of course, make one take pause and ask in hushed tones,  "what about Microsoft?" ;-)

On the other hand, I do think compatibility at the application state layer is likely a tarpit.

Let's focus on getting portability within platform "solar systems" and worry about connecting those systems once we conquered that.

I'm not quite sure what you mean by that.

Cheers

Stu

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "Cloud Computing" group.
To post to this group, send email to cloud-computing@googlegroups.com
To unsubscribe from this group, send email to cloud-computing-unsubscribe@googlegroups.com
For more options, visit this group at http://groups.google.ca/group/cloud-computing?hl=en
-~----------~----~----~----~------~----~------~--~---