"If you ellaborate on this question a bit more I could sink my denture on it."
Take, for example, the border gateway protocol, or BGP. Since it doesn't scale to accommodate the number of routes (autonomous system numbers) that are envisaged to be in place in about five to seven years, or so, it must either be radically tweaked, or replaced.
Replacing BGP, as is the case with replacing IPv4 with IPv6, would require that whatever replaces it is able to operate in a both modes, simultaneously. How many hundreds of thousands of border routers are on the Internet today that would have to be upgraded or replaced in order to achieve this?
As an aside, some folks are proposing an Optical Border Gateway Protocol, or OBGP. Bill St. Arnaud in Canada (CAN Net*4) is proposing such a scheme, whereby individual users (universities and research organizations, at first) own their own lambdas and swap them amongst themselves and neutral carrier exchanges.
This would entail every user organization having their own Internet routing domain (hence requiring each user org to employ a form of BGP, in this case OBGP... when this extends to so-hos and residentials, as they propose, can it scale, given what we know about BGP at this time?), requiring a modification, or departure, from normal approaches to end point addressing, and this is viewed as being feasible only within a "walled-garden" environment, at this time.
Similarly, GMPLS and UNI schemes (which, in some ways, are mimicked by OBGP) that are now being explored present methods of signaling between Internet nodes that run contrary to how things are done today, often using out-of-band signalling, a la SS7.
Consequently, these schemes are seen as carrier and ISP tools, and do not necessarily avail themselves to end users on the open Internet, although they could certainly be leveraged by anyone in a closed environment.
Assuming that these approaches are even desirable, getting their interfaces introduced into everyday Internet traffic flows is another example of the complexities that one would encounter when negotiating a "work around" of the Internet's legacy constructs. I'm not knocking what has been done on the net to date, mind you. Forsooth! I'm merely pointing out how we easily become limited in the ways in which we might be able to introduce improvements, once older constructs are hardened in place.
FAC |
