To: Ray Jensen who wrote (675 ) 6/2/1998 8:09:00 AM From: Frank A. Coluccio Respond to of 3178
Ray, >screaming for a better way to get wired...< Wow, that's a powerful line. Sounds like the battle cry of a militant group of bandwidth-starved fiber aspirants in the George Gilder Camp, or possibly the infamous HNCN, i.e., the Have-Not Camp of the Nineties. In my lead-in to the referenced article...Message 4667490 ... I actually misrepresented what I intended saying, for some inexplicable reason, using LA instead of Frankfurt or Rome or some other city as one of the end points in the hypothetical call, and I assumed that readers would understand what I was suggesting. In case I was unclear and readers did not adjust to what I meant, let me try it again, this time substituting Rome, Italy in the following sentence, and it may become clearer, since in this case, all roads lead to 60 Hudson Street in NY, and not Rome, as the saying goes: "How does this affect latency, when a call is placed, say, from a Rome ITSP to a UK ITSP subscriber, and it must traverse the Atlantic not once, but twice? Maybe more times, in the case where the sub is at a considerable distance from the UK ITSP's hub location, or in another country altogether? Rome--> 60H (crossing #1) 60H --> UK (crossing #2) Even under ideal conditions, this would begin to take on the delay characteristics of a satellite call, due to the mere propagation delays and transiting latencies in the router and gateway ports. If the London UK ITSP subscriber happened to be in Belfast, say, then it could conceivably make a third and fourth crossing, due to the way routing tables are set up for least cost. But one would hope that the UK ITSP would have defined routes in such an instance to prevent this from happening, as in a nailed up route definition, but I wouldn't bet on it, since this adds to their costs. On traditional POTS 64 kbps connections, or even low-bit-rate-voice alternatives, this multiple backhaul arrangement would be tenable, due to the lower latencies of the deterministic approaches used, although there are some appreciable delays introduced in high compression LBRV algorithms, but negligible in comparison to multi-hop public Internet calls or even private IP multi-hop traversals of the kind suggested here. By the way, are you going to SUPERCOM in Atlanta? Steve Guich had asked me the same question, come to think of it, and shame on me, I never got around to answering him on that one, because I've been trying to carve out the time, but thus far I'm having a hard time shaking loose some obligations. Doesn't look good at this time... [Steve, if you're there, I don't know _yet_. <s>] Regards, Frank Coluccio