From: James Ventre (messageboard@ventrefamily.com)
Date: Thu Jun 08 2006 - 15:47:03 ART
I also wanted to add one tidbit.
Unless your source/destination IP's are the same as the user traffic
(your investigating), you could be CEF'd over a different set of links
(different hash) ... further invalidating the test.
James
James Ventre wrote:
> I'm going to assume that you are not the ISP here.
>
> Suppose you have 3 segments, and 3 routers. [YOU ARE HERE] - A - B - C.
>
> If you want to to figure out the latency between A and B ... you can
> ping C. What you will get is extra latency induced from going from B to
> C and the associated process level overhead @ router C, but you know
> that your latency is no greater than that number. To make sure that
> don't send 5 pings at the wrong time (while it's getting polled), send a
> bunch and take an average.
>
> Yes, I know it sucks, but when you're dealing with punted/process level
> packets, it's never easy (especially when you don't own the gear)
>
> James
This archive was generated by hypermail 2.1.4 : Sat Jul 01 2006 - 07:57:32 ART