I think what really needs to be done is use the nfdump syntax to reduce the number of returned items. for example, what do you really want to know? Lets talk port 80. Maybe you want to know who is the biggest traffic user on port 80. From a command line try this.
Notice that I used -n 20 to only get the top 20 from the list, and -s to get source IP showing the byte count. I also used a shorter time frame.
Code: Select all
/usr/local/bin/nfdump -M '/usr/local/nagiosna/var/crmonrah03/flows/' -R . -t '2017/07/05.07:00:00-2017/07/05.12:59:59' -N -n '20' -o 'csv' -s 'srcip/bytes' 'src port 80'Is that where you wanted to go? Basically, you need to reduce the output. Limit it by time frame versus using the totality of your collected data, by how many results make sense, and by what you want to know.
I expect that a new version will deal with other performance related issues. I hope. I'm a user of the tool, we paid for it, I like it, but sometimes I want to punch it.
Thanks
Steve B