nanog mailing list archives

Re: ISP performance


From: Sean Donelan <sean () donelan com>
Date: 10 Jul 2000 13:50:37 -0700


"Well-engineered" is a loaded term, and presumes unchanging routes are
always the best way to engineer a network.  With ATM, MPLS and other
provider games the IP-level path may have very little to do with where
your packets go.

What you are looking for is the work done at MERIT and later the Skitter
project at CAIDA.  See www.caida.org for more infomation.


On Mon, 10 July 2000, Rajesh Talpade wrote:
At the risk of making a possibly naive request, could 
someone point me to data on ISP network performance 
(routing stability, packet loss, latency) ? I am 
primarily interested in the first metric, and am 
trying to understand the relative stability of large,
_well-engineered_ ISP networks. In other words, with
what degree of confidence can I predict the IP-level
path between two PoPs of a large ISP. Please note 
that I am not getting into the general Internet 
performance here.

Somehow I get this feeling that the answer would be
a fairly large...it depends ! It would be useful to
understand the influencing factors as well.





Current thread: