dmv wrote: Acidus wrote: I'm more interested in allocation of address space to these devices instead of domain names. These devices *must* have globally unique addresses. CIDR, NAT and other technologies that have slowed the consumption of IPv4 address won't help you here.
Sure they could. It just becomes a directory lookup, with roving physical addresses. Rather than give you my IP address, I give you a means of contacting me -- a phone number, a skype alias, whatever. When I establish a net connection, I register that location with my directory service, and that registration may be NAT:someopenport. They have to have globally unique addresses, but the address space does not need to intersect with the other globally unique address spaces.
It does if you want to use any IP tricks, such as IP multicast or RSVP. While some of these things have gone the way of the dodo, I have a sense that they might reappear. A glut of bandwidth caused them to fade into the background as a means of giving us quality service. But in a few years (a few short years, maybe 5), when every tivo and cell phone and xbox and refridgerator and car needs access to the network, particularly for things like streaming media, then you'll probably see a resurgence in using some of these techniques to lighten the load on the network. In order to use some of these techniques (even things like diffserv), you need to have a stable routing structure and ideally, a good solid supernet of aggregated addresses to help with the administration. In the not too distant past, I've seen highly fragmented networks collapse under their own routing chatter. They don't even need to be fragmented to generate a lot of chatter. Simply appearing and disappearing (like a mobile network would likely have) can be very stressful. For services like VoIP or VoD, where you absolutely have to have a smooth and predictable latency and throughput, having to change routes in the middle of streaming would be suboptimal. Scaling that to hundreds or even thousands of nodes would be disasterous. RE: .tel sTLD RFP Application |
