That's showing the average utilization.

Really average utilization is the only way to look at it. It just depends on
how much time you average over. Max instantaneous utilization will always
end up being 100%. Why? Because communications can only happen at wire
speed. When packets get to a router, they can be staggered out over several
layer 2 frames/cells, so that the utilization averages out to whatever is
configured or bandwidth is shared among many conversations. But when a
packet IS sent out, the only thing that can be done is to fill entire
frames/cells. So during the time that it takes to play out a frame or cell
the communication is taking 100% of the bandwidth. It might wait several
frames or cells before sending another chunk of data, and average out at
less than 100%. But during any particular INSTANT (0 time elapsed - e.g. 1
bit) the utilization will be either 100% or 0% - a bit either contains valid
data or it does not. Does that make sense?

I would suggest that MRTG (www.mrtg.org) might be a better tool for
measuring utilization - it's pretty much designed to graph this type of info
and is lighter weight if you're not worried about what's actually being sent
in the packets.

Chris 

-----Original Message-----

Date: Tue, 13 Jul 2004 09:45:56 -0700
From: "John Atkinson" <[EMAIL PROTECTED]>
Subject: [Ntop] Measuring Peak Bandwidth Use
To: <[EMAIL PROTECTED]>
Message-ID: <[EMAIL PROTECTED]>
Content-Type: text/plain; charset="us-ascii"

Hi, there.
 
My goal is to measure the peak bandwidth utilization of our T1 line, to
learn whether are needs are low enough to be serviced by a fractional T1.
 
I'm new to ntop.  I purchased and registered the pre-built Windows
installation package ( from http://www.snark.it/ntop/cart.php) and installed
it on a Windows 2000 Pro system.  I placed a simple 10/100 hub between my
firewall and my T1 router, and connected the Win2K computer to that hub,
alongside the firewall.  
 
I believe this lash-up will expose all the Internet traffic we send and
receive to ntop.  Right?
 
I played around with the start-up parameters, to try to make ntop's job
easy, given that it's running on an old Pentium 2 system with only 128MB of
RAM.  I used these start-up parameters:
 
ntop /c -z -b -n
 
I've been running ntop for a few days, and watching it from another
computer.  I focus mainly on the Network Load Statistics graph under Summary
> Network Load.
 
I notice the graph plots a data point every 120 seconds.  My question is
this:
 
Does the plotted data point represent the largest instantaneous bandwidth
utilization seen during the 120 seconds, 
 
or
 
does it represent the average bandwidth over the 120 seconds, as in
TwoMinuteTotalTrafficBitCount/120 ?
 
If it's the former, great!  It's what I need.  If it's the latter, can you
suggest how I might obtain a measurement of the highest instantaneous
bandwidth utilization we experience?  Am I even using the proper tool for
this job?
 
fyi, my router is a Cisco 1700, and is owned and managed by our T1 provider,
XO.
 
Thanks!
John Atkinson

_______________________________________________
Ntop mailing list
[EMAIL PROTECTED]
http://listgateway.unipi.it/mailman/listinfo/ntop

Reply via email to