From your and Dan Thornberg's post, I think I'm getting the idea.
Your post I think describes a more flexible solution, supporting various services via ADSL. Which is good, I was actually just thinking about the more basic problem of how ISPs will handle vanilla internet access (IP packets) on the ADSL link. The reason I'm focused on just this is that I think this will be the sweet spot of the market in the near/medium term given the web as the "killer app".
I gather then that the ip packets could be chopped up into cells by the DSLAM (is that a generic term or just westell's?), and then send them on a pvc (one for each ADSL subscriber) across the trunk to the more central point. On that more central point, many ATM links would come in (to a router?), and the packets would be routed on their way (maybe on ATM, maybe T3, whatever).
Seems like that kind of wastes the power of ATM and creates a lot of overhead, no? All that is needed I think is to pull out the ip packets from each adsl subscriber and put them on a bigger pipe (statistically multiplexed) and send them to the more central location. Then the router could decide which long-haul atm link to use.
As far as adsl much choices (see dan's post), it looks to me like the ipsilon ip switch solution in Dan's post is the ideal solution, no? Even with amati's adsl/ethernet/router solution the router isn't really needed, (there's just one direction to go, up the trunk!), just bridge/switch is needed. Isn't this a lot more cost-effective?
Regards,
Rob |
