On Thu, 27 Jun 2002, Ken Chang wrote:
> first, TCP has a curve to climb before it can explore and get right
> the available bandwidth, and this process is slow (10 secs?).
Where do you get your 10 second figure from?
> it's not the width of the channel but the stability of RTT that keeps
> the TCP climbing towards the maximum speed.
Can you explain the effect of RTT in this situation? This is not
exactly how I recall TCP working.
> DoCoMo's TCP version, TLP, performs maybe only 1/3 as the WAP stack
> in the worst case but on average maybe they can do 80%, because
> i-mode is mostly used "in a sofa-bed at home".
Well, there you go. 80% of the performance is not so awful now, is it?
> also, even in the fixed network, TCP has a lot of problems for
> streaming and signaling etc.
Would you care to describe these in detail?
> and we know most of them many years
> before UP and i-mode. but there is still no universal solution
> yet to replace TCP, a not-so-good but great design.
And who is interested in replacing TCP? What charcteristics would a
"better" protocol have?
cjs
--
Curt Sampson <cjs_at_cynic.net> +81 90 7737 2974 http://www.netbsd.org
Don't you know, in this new Dark Age, we're all light. --XTC
Received on Thu Jun 27 13:21:37 2002