[lug] Traceroute question

[Sean Reifschneider]

On Wed, Dec 13, 2000 at 01:44:39PM -0700, Michael Deck wrote:
>Why would I see the world in 200ms hops? I thought that traceroute 
>presented the inter-hop time? That was really my question...

Let's assum that you have the following hops and latencies in your connection
from the home to the remote:

	remote
	        10ms
	hop1
	        1ms
	hop2
	        1ms
	hop3
	        1ms
	[...]
	        1ms
	hop11
	      200ms
	home

So this means that data sent from remote to hop1 has a round-trip time of
10ms, and from hop1 to hop2 it's 1ms.  When you traceroute from the remote
side, it sends a packet to hop1 (10ms), then hop2 (10+1ms) then hop3
(10+1+1ms), etc...

If you traceroute from home, it first sends a packet to hop11 (200ms)
then hop10 (200+1ms), etc...  See, if it takes 200ms for data to go
from your home machine to the next hop on the internet, you can't
reach anyone in in less than 200ms.  The rates are addative.

To make it a more physical example, imagine that it takes you 4 hours to
drive from home to DIA (it's in Kansas, remember), and then 90 minutes to
fly from there to Las Vegas.  The time required for you to get from home to
Las Vegas via DIA isn't going to ever take less than 4 hours, even though
the connection from DIA to Las Vegas is really fast...

Though 200ms for the first hop on DSL seems pretty amazingly slow.  Looks
like our RADSL line is running around 40ms.  200ms is modem speed (hence
people talking about modems to you).

Sean
-- 
 Florida law dictates re-counts in the event of a close vote.  Why are Bush
 supporters rioting to break the law?
Sean Reifschneider, Inimitably Superfluous <jafo at tummy.com>
tummy.com - Linux Consulting since 1995. Qmail, KRUD, Firewalls, Python