[e2e] TCP, retransmissions, and ISPs with byte caps

Philippe Strauss philou at philou.ch
Mon Jun 17 23:17:01 PDT 2002


On Mon, Jun 17, 2002 at 10:20:06PM +0100, Jon Crowcroft wrote:
> 
> a la recherche de temps perdue

du temps perdu :)

> a single TCP
> connection in possessiion of a small fortune in source
> data
> must be in search of
> a 10Gbps fiber transcontintental link
> with an especially unusual loss characteristic
> 
> as jane austin once said
> 
> is not per packet processing
> 
> 
> its simple feature of the control loop

mice and big fat pipes?
data corruption?

> if you dont get it, go back to engineering grade school - its not
> retransmssion....

ah.. school must no be the culprit
hmm I'd rather smoke the same strain as you :)
note that TCP is doing the routing in your quest for le temps perdu :)

> sure there's lots (most) of lie
> when it doesnt happen
> 
> but when it does, its your highest paying custimer, so listen up.

here I don't get it.
I give ma langue au chat :)

But about delays, CATV systems have an interesting characeristic:
a ~20ms minimum delay in the ACK path (direction from home
to the hub, upstream or return path in catv speak).
I suspect it may lead to intersting TCP global synchronization
sometime on headends used at nearly full capacity, but
observing it is difficult.
also, since both ways of communication in a CATV system are handled
differently, you may find sometime, burst of data corruption in the
upstream channel (mostly the ACK path for a customer).
DSL also has RTT much higher than the transmission time. (due to echo
cancellation at layer 1??)

cheers.

-- 
Philippe Strauss
http://philou.ch/

L'indifférence est le plus grand risque de notre temps,
la forme civilisée de la cruauté.  -- Zenta Maurina
--




More information about the end2end-interest mailing list