Ira Abramov wrote:

Quoting [EMAIL PROTECTED], from the post of Sat, 30 Nov:

On 29-Nov-2002 David Bergman wrote:

Here's a problem that *sounded* impossible ... I almost regret posting
the story to a wide audience, because it makes a great tale over drinks

<snip snip snip - most of the story>

indeed, you could have sent a link instead of the full text.

Hear hear.

there are several points in this story that make it unbelievable in my
book. first, that 3ms delay is a lot in early 1990s standards,

a lot, or not a lot?
Early 1990s means that almost only universities are on the Internet. This makes the net itself more managable.

second
that sendmail as a user process, manages it's own TCP connection timeout
when that's a kernel thing. I'm no great programmer, but I think
creating a TCP connection is a blocked call and the timeout is fixed in the
kernel.

Hmm, not for server processes, no.
If a server program (a real one, not the kind you write as homework) used blocking calls, it'd require a process/thread for each TCP connection it created. HUGE overhead. What you actually do is use non-blocking calls. They look just like the blocking ones, only you use an ioctl to set the file descriptor to non-blocking, and then use 'select' to find out who is ready for sending, or has information been recevied. When you call select, you use a timeout. Specifically, you don't want to let idle connections hang around for too long, or someone can seriously DoS you.

furthermosre, it assumes all internet links go in straight lines
and are equaly wide and equaly congested and that the backbone is indeed
that clean.

And how much variance are you experencing today in connection setup times? And this WAS 10 years ago, when a lot less users were on the net (hence -less variance).

a physicist, ignoring the processing time of cisco
microcontrollers and the delay of repeating packets from one side of a
switch to the other, and then from ether to WAN and through a modem,
would STILL argue that electricity doesn't travel at the speed of light
either.

No, they won't. They will claim that the ELECTRONS don't travel at the speed of light, but that is mostly irelevant to this discussion. The speed of electricity is not far from the speed of light, IIRC.

and there are a few other problems.

Do tell.


in short, only a cute suburban legend :)



=================================================================
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]

Reply via email to