[PATCH] Added NoDelay config option and nodelay subsystem option

Rick Jones rick_jones2 at hp.com
Wed Jan 30 07:01:31 EST 2002


 
> > Did I get it right, or did I miss something?  As you can see,
> > transmitting four data packets took 4.4 seconds.  Without Nagle, it
> > would have taken 4 second.
> 
> yes, 200ms is way too much nowadays. Which is why Nagle needs
> updating, and why Nagle could be done much better at the SSHv2
> layer...

My highschool had an excellent Physics teacher who knew how to make a
lasting impression. At the beginning of the term, he wanted to make
sure that all his students understood the distiction between mass and
weight.  So, he had all of us stand-up and start jumping up and down,
shouting, at the top of our lungs, "Mass is NOT Weight! Mass is NOT
Weight!" (imagine 20 or so prep-school boys in coats and ties if you
care to complete the picture :)
 
So, in this context, it would be good for everyone to stand-up at your
computer and start jumping up and down and "shouting (quietly so as not
to disturb the neighbors)

 "Nagle is not Delayed ACK!"
 "Nagle is not Delayed ACK!"
 "Nagle is not Delayed ACK!"

Indeed, Nagle and Delayed ACK are related (perhaps not as intimately as
mass and weight but it was the best I have at the moment :) but they are
not the same.

Also, I know of at least one stack that uses a 50 ms standalone ACK
timer :) So, in the example, it would have been a difference of 4.1
versus 4.0 seconds so a 2.5% difference rather than a 10% difference in
clock time.

rick jones
-- 
Wisdom Teeth are impacted, people are affected by the effects of events.
these opinions are mine, all mine; HP might not want them anyway... :)
feel free to post, OR email to raj in cup.hp.com  but NOT BOTH...



More information about the openssh-unix-dev mailing list