Need help on process netflow sampling data

Joe Shen joe_hznm at yahoo.com.sg
Wed Nov 24 14:37:49 UTC 2004


Hi,

I'm trying to analize our egree router traffic by
using flow-tools and CUFlow.

There are three edge routers: two Juniper M160 and one
Cisco GSR. All of them are set up to sample outgoing
interfaces.

With Juniper M160, I set up forwarding-option as:

=============================
forwarding-options {
    sampling {
        input {
            family inet {
                rate 500;
            }
        }
        output {
            cflowd 10.101.172.20 {
                port 2055;
                version 5;
            }
        }
    }
}

 so-6/0/0 {
        description "Outgoing 2.5G POS";
        unit 0 {
            family inet {
                filter {
                    input netflow-sj;
                    output netflow-sj;
                }
                address *.*.*.110/30;
            }
        }
    }
====================

With Cisco GSR I configed:

======================
ip flow-export source Loopback0
ip flow-export version 5 origin-as
ip flow-export destination 10.101.172.20 2055
ip flow-sampling-mode packet-interval 500
ip flow-aggregation cache as

interface GigabitEthernet5/0
 description To_Other-AS
 ip address 10.96.117.250 255.255.255.252
 no ip directed-broadcast
 ip route-cache flow sampled

====================

To my understanding, they should work the same way as
: sampling every 500 packets in both in & out
directions.  I set up CUFlow.cf to work with Sampling
rate 500, but when I try to graph data with its cgi
scripts, i found CUFlow show a wrong bit rate with GSR
( 10Gbps for two GE ), while there is much differnt in
application analysis. 

Is my understanding of "sampled netflow" correct?
Is there anybody could do some help with netflow data
processing ?

each word will be highly appreciated


Regards

Joe



__________________________________________________
Do You Yahoo!?
Log on to Messenger with your mobile phone!
http://sg.messenger.yahoo.com



More information about the NANOG mailing list