anyone know what the canopy packet per second is? or however they measure it. i have a ton of arp traffic, i know most of it should be normal but, what is an acceptable arp count for 800 customers?
vince
im getting over 1000 arp requests a minute
i’ve got a couple of threads open on this pps:—
arp is a nightmare…
an SM can send upto 300pps to an AP
an AP can send upto 1500pps to an SM
an AP can receive and send upto 3000pps
a BH can handle 3000pps
The CMM (standard switch MIB) can be SNMP’d for that info, but canopy to the best of my knowledge to not give that info, even though the pps is the real limiting factor of the kit and not the bandwidth.
using ethereal you can see who is generating the arp traffic and then you should be able to calculate the path and see how pps it is generating. Also arp is broadcast so you should be able to just take 1000/60 = 16.67 pps. What is your setup and where are you picking up your stats from.
Just be careful of broadcast storms, from what I have read occur due to IGMP traffic hitting a NAT disabled SM, causing it to trip
my set up is…
orthogon BH to main site then all sites BH’ed from the main site
I have the ds3 router and servers in the office where the orthogon BH feeds from.
I have a port on my cisco router mirrored from the main orthogon port to another port that i sniff from.
between tcp/ping/arp/dhcp/icmp im getting an insane amount of traffic. my ethereal files are 100+ megs just sniffing for a minute or two.
ICMP replys still manage to own me…sure i can track down the IGMP request but it just takes time and understanding customers. Im in the process of splitting my network up into 8 different pieces to help minimize damage.
If i would of known about this problem i would have all my SM’s nat’ed but i already have 800 customers with routers that may or maynot be configured correctly.
these customers are all nat disabled ?
how do you manage their network addresses… their routers are configured by you ?
you can run nat enabled with DHCP server running…
should start switching them one by one…