On Sun, 16 May 1999 22:32:55 +1000 (EST) 
 Darren Reed <darrenr@reed.wattle.id.au> wrote:

 > How about using high/low watermarks for "despiration" ?  These variables
 > would be sysctl'able, of course.

Then you would also need a variable for the level at which each protocol
did the drain.  This could get complicated :-)

 > IMHO, the draining routines should also take into account how long a
 > packet has been waiting in the queue - that is, expire the oldest first,
 > in an FIFO type schedule.

Actually, I would say start by dropping th youngest packets, so as to be
more like if they were dropped at the h/w interface level due to starvation.

This would also tend to preserve the order of packets, which might make
applications happier.

        -- Jason R. Thorpe <thorpej@nas.nasa.gov>