Re: Variable Link Delay

[John Heidemann <johnh@ISI.EDU>]

On Tue, 14 Sep 1999 14:33:19 CDT, Manish Airy wrote: 
>I'm trying to model the effect of a variable link delay (given an
>empirical delay distribution) on TCP.  I managed to change the recv
>method of class LinkDelay to include the delay statistics I need.  This
>works fine if there was only one link in the simulation.  However, not
>all links in the simulation need to follow this distribution.  Is there
>a better way of doing this so that different links have different
>empirical delay distributions  ?

How about subclassing LinkDelay?  Your changes then go in the
subclass, and there's some new Tcl API to create links with your
kind of delay rather than the standard one.

This would be a useful thing to have.

   -John Heidemann