Subject:
|
Re: "jump.cgi" considered harmful ? (1)
|
Newsgroups:
|
lugnet.off-topic.geek
|
Date:
|
Fri, 1 Dec 2000 21:19:26 GMT
|
Viewed:
|
520 times
|
| |
|
|
In lugnet.off-topic.geek, Larry Pieniazek writes:
> OK, well I am confused now. I thought that since tracert can't really ask
> all the routers on the routing to report interroutter times, that what it
> displays is the time from where the trace is run to that particular router,
> one router after another... that is
>
> the time from where I am to 10 is 10,10,10 MS over three tries
>
> the time from where I am to 11 is 10,10,10 MS
>
> ...so probably almost instantaneous but you can't be sure because this
> tracert actually ran different tests to get the node 10 numbers than it did
> to get the node 11 numbers
>
> the time from where I am to 12 is 80,90,90
>
> ... 12 is the first node on the other side of the atlantic.. or that it
> takes 70-80ms to cross the atlantic (5-7 times the speed of light time for
> that distance, more or less, so not that unreasonable)
>
> and the times to 13, 14, 15, etc don't go up much (implying short node to
> node times) till we get to the cais nodes...
>
> or am I misunderstanding how tracert works?
No, you're right. I don't understand how you're confused.
If it was my original assertion that it might have been the UUNet network that
was causing the suck, well...
You got how traceroute works completely right. But something to keep in mind
is that packets that go out to the higher steps come BACK through the lower
steps. In other words, packets going to 13, 14, etc go through 12 not once,
but twice.
Now, keep in mind that these are ICMP packets, which are absolutely the first
packets that get dumped in a congestion situation.
Since I know that TCO, like many UUNet peering hubs, has a tendency to get
congested, it's easy to jump to the conclusion that congestion at TCO was
causing later packets to get dropped, and therefore time out, even though the
one reaching TCO actually reached TCO's serial port just fine. If it's another
serial port that's congested on TCO, and that serial port happens to be the one
that you need to get to your destination, that is exactly what you'll see.
Especially if the processor on the router in question isn't having a problem.
Anyway, I was diagnosing in stream of consciousness back there. Cutting UUNet
out completely proves to my satisfaction that the problem was defintely with
cais.net's network, and not UUNet's. This time.
eric
|
|
Message is in Reply To:
 | | Re: "jump.cgi" considered harmful ? (1)
|
| (...) OK, well I am confused now. I thought that since tracert can't really ask all the routers on the routing to report interroutter times, that what it displays is the time from where the trace is run to that particular router, one router after (...) (24 years ago, 1-Dec-00, to lugnet.off-topic.geek)
|
36 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
