the promised text


Good observation. However, I did not miss Vadim's point. It is well
stated. But I would like to see a clearer path to deterministic behaviour
in routing technology before we use it for integrated voice service, as an
example. "No worse than ATM" is not a good enough starting point.

Indeed, today we use ATM to transport IP traffic and also multiple
technologies including some non-ATM stuff like broadcast video/voice, laser
disk playback, Video Conferencing, and ordinary T1 AMI/D4 PSTN voice
service. I wonder, however, if an IP address (even at terabit speeds) will
help to sort out which DS0 is going to which IXC. Likewise, we should never
think that IP traffic will ever be limited to any specific L1/L2 technology.
Because I can do these things with ATM today does not mean that better IP
routing engines are not needed!

I mean no disrespect to Vadim. I think is work is good. I actually hope we
do see better IP routing technology. If the resultant product is
deterministic at L3, ATMnet would probably be one of the first to sign up to
give it a try.