Let's remember that a major goal of these facilities is to get a user to a
server that is 'close' to the user. Having interception done only at
distant, localized server farm facilities will not achieve that goal.
granted, but...
an interception proxy that gets the user to a server that is 'close'
to that user (in the sense of network proximity), but 'distant' from the
content provider (in the sense that it has a significant chance of
misrepresenting or damaging the content) is of dubious value.
and a technology that only works correctly on the server end seems
like a matter for the server's network rather than the public
Internet - and therefore not something which should be standardized by IETF.
I do think there is potential for standardizing content replication
and the location of nearby servers which act on behalf of the content
provider (with their explicit authorization, change-control, etc).
But IP-layer interception has some fairly significant limitations
for this application. For one thing, different kinds of content on
the same server often have different consistency requirements,
which become significant when your replicas are topologically distant
from one another. If you treat an entire server as having a single
IP address you probably don't get the granularity you need to implement
efficient replication - you may spend more effort keeping your replicas
consistent (and propagating the necessary state from one to another)
than you save by replicating the content in the first place. Obviously
you can use multiple IP addresses, assigning different addresses to
different kinds of content, but this also has limitations. You can also
get into problems when the network routing changes during a session or
when the client itself is mobile.
Bottom line is that IP-layer interception - even when done "right" -
has fairly limited applicability for location of nearby content.
Though the technique is so widely mis-applied that it might still be
useful to define what "right" means.
Keith