[30353] in North American Network Operators' Group
SUMMARY: bw usage?
daemon@ATHENA.MIT.EDU (David M. Ramsey)
Fri Jul 28 09:49:23 2000
From: "David M. Ramsey" <dmr@webserve.net>
Message-Id: <200007281347.e6SDlSg07089@prizm.webserve.net>
To: nanog@merit.edu
Date: Fri, 28 Jul 2000 09:47:28 -0400 (EDT)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Errors-To: owner-nanog-outgoing@merit.edu
Thanks to everyone who shared thoughts, ideas and experience
regarding monitoring bandwidth usage on Ethernet switch ports
without including broadcast traffic.
EVERYONE tends to agree that a separate VLAN for each co-located
customer is the only professional way to do things "right".
I agree, and plan to move in that direction, fast. Then I won't
have to worry about broadcast traffic, either!
Other recommended solutions included:
- MRTG
This is great (I use it elsewhere) but it doesn't directly
address the issue I have of *not* including broadcast traffic
- Cisco 6500 switches apparently support "Private VLANS", which
don't burn up IP addresses. Sounds cool, wish I had a 6500 ;-)
- AUI port -- several people pointed out that an unused AUI port
on a switch will log broadcast traffic and no other; I can subtract
that from traffic monitored on all other ports. Some people
said that an unused VLAN1 has same behavior.
- dumb hub -- by plugging in a dumb, passive Ethernet hub with
nothing connected, it will receive broadcast traffic. I can
then subtract that ports byte counters from others.
- Cisco NetFlow / cflowd ( http://www.caida.org/tools/measurement/cflowd/ )
Seems very thorough, overkill for me right now.
- RMON counting of unicast-Octets only
- Extracting SNMP # multicast/broadcast packets, using average packet
size and subtracting product from total byte count
- HPOV + SNMPv2
Regards to all, --dmr
David Ramsey
Charlotte, NC
