Right church, wrong pew.  It's actually a smidge more complex - we report
traffic based on the lower port # we recognize.

So if we come in in the middle and see 80->32451, we call it HTTP(80) same
as if we saw the initial 32451->80 request.


For me, Plans = Supported Development.  $0 supported development = no plans.


Remember:

Layer 3 (e.g. http) are handled via the -P protocol list.

Layer 2 protocols are more complex - that takes coding.

Similarly, deep inspection (into the packets of a layer 3 protocol) takes:
   (1) coding
   (2) access to packet contents - NetFlow?  Buzzz but thanks for playing...
   (3) a reality check

(WRT #3 - we don't, for example, want to maintain a site wide connection
tracking table to monitor - say - NAT or P2P sessions)


-----Burton


-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Gary
Gatten
Sent: Thursday, May 11, 2006 3:06 PM
To: [EMAIL PROTECTED]; [email protected]
Subject: RE: [Ntop] erroneous service reporting

I guess that's what I was saying.  AFAIK you (nTop) do simple port matching
right?  No deep packet inspection or protocol analysis?  So if Bittorent (or
anything else) uses 554, it will show up as rtsp.

Is there any plugins or planned support for deeper protocol analysis for
more accurate traffic stats?  I've done some stuff with Cisco NBAR and
policy based routing / NAT so nTop always see's the same port for a specific
application - regardless of what that application tries to use.
 But, that only works in certain configs.  Would be nice if nTop could look
deeper to determine what's really using a specific port.  Of course, with
netflow I don't think this would work at all would it? 
Since you don't actually get the packets when using netflow, there'd be
nothing for you to analyze.  I THINK netflow "can" report addy/ports post
NAT processes, so my NBAR/ PBR thing may actually work. 
Interesting...



>>> [EMAIL PROTECTED] 5/11/2006 2:50 PM >>>
Um...

  Real Time Streaming Protocol (RTSP) (RFC 2326 and others I imagine) define
the standard port as 554. Bittorrent defaults to 6881, but that's almost
always changed...

Still, I don't see the collision...

-----Burton



 

-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Gary
Gatten
Sent: Thursday, May 11, 2006 2:14 PM
To: [EMAIL PROTECTED]; [email protected]
Subject: RE: [Ntop] erroneous service reporting

I think he means Real Time Streaming Protocol.  I think your explanation
still holds true however.

Gary


>>> [EMAIL PROTECTED] 5/11/2006 2:04 PM >>>
RSTP is layer 2 (Rapid Spanning Tree Protocol).  Bittorrent is based on
tcp/ip (protocol 17).  Or do you mean some other RSTP?

If this is all layer 3, then read docs/FAQ - the article on how ntop
classifies traffic.

You may need to remove protocols from the list if they aren't on your
network there's also an article on that.

-----Burton
 

-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of John
Fountain
Sent: Thursday, May 11, 2006 1:14 PM
To: [email protected]
Subject: [Ntop] erroneous service reporting

List-

We are having an issue where we believe ntop may be reporting rtsp streaming
traffic as bittorrent. Are there any known issues regarding this situation?

thanks

-john
_______________________________________________
Ntop mailing list
[email protected]
http://listgateway.unipi.it/mailman/listinfo/ntop 

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


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

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


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

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

Reply via email to