Cell Relay Archive

[cheung@mathcs.emory.edu (Shun Yan Cheung)]

In article <3n969h$1mh@prop.wes.army.mil>,
Scott D. Weberg <sweberg@wes.army.mil> wrote:
>
>I keep reading that the TCP window size has a lot to do with how
>fast you can push data thru a unix socket over ATM.  

The window size can also determine how much you need to retransmit
when the head of the packet is lost :-).

>But I still don't know what the "TCP window size" is - and how it is
>set.  There seems to be no socket routines that allow you to set
>the TCP window size.  Can someone tell me how you set it?  Is it
>something the sys admin has to do?

Since Van Jacobson published his classic flow control SIGCOMM work
(check in SIGCOMM'88), TCP has been using a dynamic window scheme 
(the scheme is very similar in flavor as one developed by Raj Jain 
when he was at DEC).

Basically, TCP always starts with a window of 1, climbs exponentially
to a max (negotiated at call setup) and then climbs linearly until
it starts loosing packets. When it looses one packet, then TCP cuts 
the window in half and resume linear climbing (fast retransmit).
If it looses multiple packets, TCP starts from scratch (window = 1)
after setting max. size to half the current size.

-- 
Shun Yan Cheung     |  cheung@mathcs.emory.edu      | Office: (404) 727-3823
Emory University    |  cheung@emory.bitnet          | Fax:    (404) 727-5611
Dept of Math and CS |  -----------------------------+-----------------------
Atlanta, GA 30322   |  ``Nee heb je, ja kun je krijgen''