I wondered what the best option would be for logging and generating reports on all internet traffic not just web traffic from within an office. I do not need to block any traffic but is purely for reporting.
Some Information I would like to capture would be:-
HTTP / HTTPS domains visited and by who
Protocol usage
The ability to view Torrent traffic especially to see who and when
I’ve tried playing with ntopng using traffic-flow however it seems a little inaccurate.
What options are recommended? If external server is required for collecting information I would rather it Linux based than windows.
traffic-flow is the way to go. what is inaccurate about it?
of course you could also setup a port mirroring or packet sniff streaming and send all traffic to an external computer doing the work.
So lets say I start a download of a 512MB file with wget on a laptop:-
wget http://ipv4.download.thinkbroadband.com/512MB.zip
--2016-10-07 22:20:36-- http://ipv4.download.thinkbroadband.com/512MB.zip
Resolving ipv4.download.thinkbroadband.com... 80.249.99.148
Connecting to ipv4.download.thinkbroadband.com|80.249.99.148|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 536870912 (512M) [application/zip]
Saving to: ‘512MB.zip.1’
512MB.zip.1 100%[===========================================================================================================>] 512.00M 16.3MB/s in 35s
2016-10-07 22:21:11 (14.8 MB/s) - ‘512MB.zip.1’ saved [536870912/536870912]
As you can see it did “16.3MB/s in 35s”. I then search for the host in ntopng, give it 10 minutes or so and the traffic counters for the host have barely registered anything, maybe a few MB. Also the speedo at the bottom that shows the overall speed fluctuates a lot but generally stays below 200 Kbps even when downloading at 16MB/s.
I must be missing something?
As for setting up a port mirror, the server I’m using for ntopng is a KVM VM so that would be quite hard.
Ok, so it appears that uploads from the client are working fine but downloads are not and the percentage between sent and received it like 98%/2% in favour of sent.
Could something like fasttrack be the issue? Could traffic incoming from the internet be skipping the traffic-flow capture?