On Friday 29 November 2002 01:21, nitin panjwani wrote:
> Hi All,
>
> We are doing some experiment, where we are trying to
> measure the processing time taken by a node when a
> packate pssess throgh it.
>
> --------eth0[Linux-PC Router]eth1-------
>
> We run tethereal within this PC at eth0(incomming
> interface) and eth1(outgoing interface) . Then we find
> out our packate in both the captures and the
> difference of two captures gives processing time of
> the node
>
> Now, the question: Is doing so a right way to measure
> the processing time, as we are not sure where exactly
> does libcap captures the packates. Is it right before
> network driver or after it ao anywhere else?
>
> I just came to know that On Linux, libpcap captures
> packets using PF_PACKET sockets (or, on 2.0
> kernels or with libpcap built on a 2.0-kernel system,
> PF_INET/SOCK_PACKET sockets).
>
> Any help on this issue will be highly appreciable.
Why not using 2 extra boxes. If you send a packet from left to right and the
right hosts responds immediate (ping or so), so the packet returns, you know
the processing time.
I think there are also traffic generator tools that can monitor delays and
jitter.
Stef
--
stef.coene@docum.org
"Using Linux as bandwidth manager"
http://www.docum.org/
#lartc @ irc.oftc.net
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
