[3659] in Athena Bugs
Re: mailhub hosedness
daemon@ATHENA.MIT.EDU (Ken Raeburn)
Wed Nov 15 23:27:53 1989
Date: Wed, 15 Nov 89 23:27:32 -0500
From: Ken Raeburn <Raeburn@ATHENA.MIT.EDU>
To: bugs@ATHENA.MIT.EDU
In-Reply-To: Ken Raeburn's message of Wed, 15 Nov 89 18:28:24 -0500,
A point I omitted: Since addresses are generally allocated starting
with low numbers, the delays (low byte of address times 7, in seconds)
tend to the small end. If we add some more hair (say, multiply by 49
instead of 7, and take remainder mod 1800 -- maybe some other number
near 1800/N that is relatively prime to 1800, for some moderately
large N that serves as an estimate of number of workstations per
subnet).
Another thing we could do is factor in the other components of the
address. (The 18 will be constant, and the third octet will almost
always be 0, but the subnet number will vary.)
It isn't expensive to do this, since all the computation can be done
within awk. The toughest part is figuring out good numbers....
