Re: [vulture@aoi.dyndns.org: Bug#100744: Binary should be in /usr/bin, since it's useful to non-admins.]
On 06/15/2001 07:10:17 AM Manfred Wassmann wrote:
>> Network load. Keeping users from running expensive traceroute probes
when
>> a simple ping would suffice.
I realize things are more expensive in Europe, but this is a bit silly.
Assuming typical midwestern USA costs, a quick rough estimate of costs of a
traceroute are:
About a millionth of a cent for network access. (figuring a full T1 is a
couple hundred GB for a couple thousand a month)
About a hundredth of a cent for electric power for the PC and monitor, desk
lamp, and HVAC (unless you type in the dark, or outside).
About 25 cents for labor cost to pay someone to watch the monitor and
comprehend the results.
So as a rough estimate, in the midwestern USA, the labor cost of paying
someone to run and comprehend a traceroute is perhaps twenty five million
times the cost of the network bandwidth to run the traceroute. Or another
way of looking at it, is the cost of the electricity to run the PC while
doing a traceroute, is about ten thousand times more than the cost of the
network bandwidth.
Obviously, a DSL line or cable modem are cheaper, power and labor are
vastly more expensive in California, etc. That only increases the ratios.
I suppose we'll have to move IRC clients that allow ANSI color into sbin,
because ANSI colors, in addition to being annoying, use more bandwidth than
someone running plain B&W text IRC.
Also I'd argue traceroute causes less network load than web browsing, so
we'd be better off with a dozen people running traceroute than a dozen
people running netscape. I guess netscape goes into sbin.
If the best arguement the loyal opposition can come up with, is that making
it harder for users to traceroute will save bandwidth, that's pretty bad.
Reply to: