[LINK] The PLAN, and broadband speeds?

[Geoff Huston]

Stewart Fist wrote:
> Geoff's point about satellite delays, only apply to roughly symmetrical
> interactive services, where the back-channel is also over the satellite.
> One way broadcast doesn't suffer delay problems.

Actually, my point applied to the TCP protocol. If you want _reliable_ 
data transfer then the peak performance of the network depends 
_critically_ on the delay, the bit error rate and the jitter, 
irrespective of the "raw" available bandwidth.

It takes a little over 160 ms to get a signal from the surface of the 
earth to a spacecraft in geostationary orbit, and another 160ms to get 
back down again. So you get around 320ms of delay. Thats a problem if 
you want to run TCP across the connection and run it _fast_

   Geoff