Re: nastygram from xxx.lanl.gov
Daniel T. Martin (MARTIND@carleton.edu)
Tue, 09 Jul 1996 08:04:13 -0500 (CDT)
I would propose that the lag time between requests be determined as some
multiple of how long the previous request took. First decide how much of
the web time of a particular server you wish to occupy during your
experiment - I would suggest less than 10%, even for a short-lived
experiment such as this one. Then simply queue your requests so that, for
the 10% figure used above, you wait nine times as long between requests as
the length of time for the previous request.