Subject: Re: TCP limited transmit
To: None <>
From: venkat venkatsubra <>
List: tech-net
Date: 11/30/2000 09:46:57

Mark Allman wrote:

> > If i wanted to avoid adding the new field t_extra, won't the
> > following  work too ?
> Um.  I'm still thinking about it.  I am not sure that it is the
> same...  And, I am thinking of other's comments.

Let me know if you find any difference.
Also,  i prefer the bumping the cwnd up, calling tcp_output
and then restoring it back method because it avoids adding code
in tcp_output that would be executed for every outgoing
packet if i understood it correctly.

> > Also, is there a simple way to count the number of times the
> > 'Limited Transmit' code helped avoid a retransmit timeout ? That
> > is, helped a fast retransmit to occur when otherwise it would have
> > resulted in a retransmit timeout ?
> As a heuristic, fast retransmit when cwnd is less than four packets
> would qualify.

This would work well unless the reason for less than four packet
cwnd is the arrival of a ICMP source quench just before the Fast
Retransmit. In which case the cwnd is reset to 1 segment.

>  Hmm...  Due to my current state of mind I can't come
> up with a concrete algorithm, but I would guess that you could
> cleverly count ACKs and figure it out.  Why?  Are you thinking of a
> counter or something?

Yes, i thought a counter that can be displayed through 'netstat -s'
could be useful.

> Thanks,
> allman