I would encourage a discussion on what overall services and
architectures we are looking at for the future of the Internet. I see
mainly three camps, the ones being interested in situations where the
ISP is in strong control (and responsibilities) over quality,
addressing, services etc. I normally call these "vertically integrated
solutions". The second group are the one-to-many applications, where
users explicitly or implicitly connect to a server from where they get
information they want. These are the Googles, Spotify, Facebook, but
also many p2p solutions that have self-elected supernodes acting as
The third group which I think is the most interesting one include the
many-to-many applications. Where end users really try to connect to
their own computer at home when they travel. When two persons really
try to communicate with each other without any broker. With some very
limited rendezvous mechanism, like DNS or other naming services, but
no broker/rendezvous on the application layer. Almost only Apple is
trying to do these things I claim. But I really really really hope I
am wrong as I want to see more of this myself.
Will the future ip-layer handle these many to many applications, or
will this be resolved by having even more overlay networks on top of
HTTP or something similar, where these overlay networks are built
automatically using supernodes?
There are both scaling and addressing issues in here. On many layers
in the architecture.
Ietf mailing list