What eric is refering to is a couple different items.  One is the forward
lookup of the name given on the command prompt, which I don't recall any
traceroute implementations which cause high latency for that.
Secondly is the reverse lookup many traceroute's will do if you give an IP
address as the destination.  Many of these send the first packet out, then
make a call for reverse lookup.    Sun Solaris is the notable OS who does
this with ping and causes the first response(s) to be reported as  extremely
high latency due to the program waiting on the reverse lookup to finish.
3rd is the reverse lookup of individual hops as seen in traceroute output.
I can't recall any implementation mangling RTT results due to this, but I
wouldn't be surprised to see it.  Mostly this just delays the next round
packets from being sent.
Finally kernel level ICMP rate limiting has been done in a number of OS's
and makes agressive ping tests a poor tool.  And makes using low rate ping
against a busy host something to trust with skepitism.

I doubt you are seeing any of these Mike, but just wanted to clarify why
someone would see those kinds of results.  I know I've had to have long
conversations explaining these things to *nix admins who believed the
network had extremely high latency.  :-)

There is obviously something going on, not sure what it is myself.  I agree
with the other posters that L2 could be causing performance problems.  Have
you broken down testing so it's not just end-to-end between these two
windows hosts but also from one windows host to each of the endpoints along
the way?  Has IKE finished already when you send these packets?  Are the
lifetimes of your SA's long enough or are they aging out between individual
test packets?

Darrell

""Mike Sweeney""  wrote in message
[EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
> In answer to Eric, there is not any DNS involved as the traceroute is IP
> only... no name resolution needed.
>
> In answer Ed's comments, I have both plugged into a switch and so it's not
> *back to back* in the normal sense of the word.
>
> MikeS




Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=61235&t=60981
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]

Reply via email to