Re: [ogpx] the term 'grid'

[Dave CROCKER <[e-mail]>]

> So the "grid" in Open Grid Protocol is accurate; it is a protocol
> intended to be deployed on a grid.

Yes and no.  It is a particular instance of the construct.  Except that the term
is really tied to an implementation model, whereas OGPX is quite explicitly not.
  So while it probably really is a legitimate instance, there is a plausible
argument that it isn't.

Fundamentally, cloud computing is remote shared service hosting.  A
long-standing, low-level service that comes into vogue every 15 years or so.
This is vastly different than the semantics of "virtual worlds".

That is, the term is tied to details that seem not to apply here.  The question
is whether it might cause confusion in folks in the general Internet technical
community.

I tend to be in favor of labels that work hard to be correctly meaningful and
relevant to outsiders and to avoid confusion. It is usually a challenge to get
the right balance.  I initially thought that the choice of 'grid' was quite
quite a good choice, until coming across its use in the cloud world.

There is a tendency to use labels that are refer to a  mechanical point at too
low a level or to be too general.  The fact that the cloud world not only uses
'grid' but apparently has some confusion about its use is what prompted me to
raise the flag.


> i'm not sure this rises to the definition of a "precise meaning."
> instead, it seems like there is general agreement that "grid
> computing" involves a loosely coupled collection of distributed hosts
> working together on the same or related problems. this is a good
> definition of the architecture we explicitly support with OGP.
> 
> also... historically... the term "grid" has been used among the OGP
> interoperability investigators to define the collection of distributed
> systems which implement the virtual world under simulation.


Exactly the issue:  It invokes implementation rather than semantic concepts.

In looking at the the postings offering alternative suggestions, I was drawn to
'Region' since it is a core term already in the existing documents, and it
refers to the entity/location that instantiates a world itself.  The word
"Region" does not have any substantial history in this realm of computing or
networking; so collisions and confusion won't happen.

If I am understanding the existing documents and the goal of the current effort,
it is targeting access to regions.  Region Access Protocol (RAP) seems like a
plausible choice.  No doubt there are others.

d/
-- 

   Dave Crocker
   Brandenburg InternetWorking
   bbiw.net