Thanks very much Burton, that was a very helpful reply.

I've decided to go with a port span.  Our traffic is low (a couple
Mbps), so the compression of netFlow isn't needed.  Also, I like the
deep packet inspection capabilities.

And yes, I made sure my monitor interface does not have an IP.  

Ntop was up and running quickly.  I'm sure more questions to follow
though....thanks again!
Shane

On 4/14/05, Burton Strauss <[EMAIL PROTECTED]> wrote:
> The advantage of netFlow is the implicit compression - 15 (or more or less)
> packets in one 1500 byte packet vs. 15 packets each of whatever length.  The
> disadvantage is that you lose the layer 2 (Ethernet) and internal (deep
> packet inspection) details.  That's a trade-off decision that depends on
> what you need to pull out of the data.
> 
> There are two other ways you can grab the data, vs. span.
> 
> (1) A passive tap.  It's easy to build for 10- and 100-BaseT (instructions
> are at snort.org).  Just remember that you will need TWO interfaces on the
> ntop box (one for each direction) and you MUST merge the traffic (the
> default, but this means stay away from netFlow!).
> 
> (2) A true hub.  Not a switching hub, but a true hub.  These aren't easy to
> find, but older 3Coms and Linksyses work great.
> 
> Span too, will certainly work.
> 
> In either case, remember you don't need to assign (*and don't want to) an IP
> address to the monitoring interface(s).
> 
> -----Burton
_______________________________________________
Ntop mailing list
[email protected]
http://listgateway.unipi.it/mailman/listinfo/ntop

Reply via email to