Wireshark-users: Re: [Wireshark-users] Link quality tests/analysis?
From: Sake Blok <sake@xxxxxxxxxx>
Date: Thu, 29 Jan 2009 06:22:49 +0100
On Wed, Jan 28, 2009 at 07:39:38PM -0500, rkruz@xxxxxxx wrote:
> thats some great insight.  Thanks!
> I had no idea errored frames were not mirrored.  Would inserting a 
> "hub" in the link and then to Wireshark allow me to see the errored frames.?

Depends on the hub, but putting a hub in between would turn your link
from full-duplex to half-duplex which will off course severly influence
the throughput you can achieve on it...

A "tap" however would sit in between passively. But then you would need
two NIC's on your capturing device (one for each direction), or the tap
must combine them already for you. And last but not least, you will need
to use NIC's that are able to pass the error frames to the networking
stack, which most NIC's won't do.

So all-in-all I would say the link counters on the device that
terminates the fiber would be the best place to measure the error rate
of your link.

> on RTT.  There is a filter for send and response to a message from  
> an IP address.  WIth ping, couldnt I filter just the ping send/rcv 
> data, export to a spreadsheet and determine the RTT by subtracting 
> the times of send and receive?

Of course you could walk that path, but why not take the shortcut with
the other tools mentioned (pingplotter etc)?

Cheers,
    Sake