[e2e] Fwd: Camel's nose in the tent

[Graham Cope]

David,

Thank-you for the answer. 

Next time I try to design an IP QoS network and wonder where to place
the MPLS Edge, Diffserv Boundary/Edge nodes and Access Servers, and
which intellegince / control features to place within them, I shall now
adhere to the definitions that you supplied in the hope that everything
becomes obvious.

(extracting tongue from cheek)

As you say though, it's a design principle operating at a layer of
abstraction, and hence some serious design work is required to map that
on to a physical realisation.


Graham





"David P. Reed" wrote:
> 
> At 09:11 AM 8/13/01 +0100, Graham Cope wrote:
> >My favourite is the somewhat glib statement used by some people
> >'intelligence is moving to the edge of the network'. I have three
> >problems with that statement:
> >1/ What is intelligence?
> >2/ What and where is an edge?
> >3/ What is a network? E.g., tunneling through, confusion between a
> >service and a network (what is the PSTN?, or in this case, what is an
> >e-mail network), virtual network, overlay (underlay?) networks etc.
> 
> It's glib, but no more so than many sensible statements.  Since you ask:
> 
> 1/ Intelligence is computational decision-making on the behalf of users or
> intentionally allied groups of users.  (e.g. error recovery involves
> decisions on behalf of users, as does security, as does "quality of
> service" in its many guises, as does "fair resource allocation").
> 
> 2/ Edges are the areas or functional components of the network under more
> direct control of individual users or intentionally allied groups of users.
> (e.g. computers under the users' control, applications, etc. as opposed to
> shared facilities operated by third parties that are not owned or
> controlled by individual users or allied groups).
> 
> 3/ A network is an interconnection scheme that facilitates communications
> among users or intentionally allied groups of users.  (e.g. the IPv4
> transport interconnection commonly called the Internet, or the HTTP
> transport interconnection commonly called the Internet - different
> abstraction, same name -, the telephony service network commonly called the
> international telephone service or POTS).
> 
> That this set of definitions has terms that can be bound at different
> levels of abstraction is a feature - because there is much that can be said
> about design principles that is independent of binding of the definitions.
> 
> - David
> --------------------------------------------
> WWW Page: http://www.reed.com/dpr.html