Hello everybody,

I'm trying to write a nfsen backend script to detect (automatically) denial-of-service attacks.

From my experience, the types of DOS attacks that I have registered against my network are these:

   * flooding from one source to one destination (seen like a large
     number of pps from the same flow)
               + these are easy to catch - I've set up a top 200 pps
                 and I compare the duration and number of pps of each
                 top result and see if they exceed a limit.
+ -n 200 -s record/pps + 'duration > 1000' as a filter (to get rid of false
                 positives - flows that have low number of packets (30)
                 and low duration (0.001s) have 30.000pps!)
   * flooding from one source to one destination with low pps
               + these have a great flows count
               + are quite rare, in my case.
               + they are caught the same way as the first type of flooding
               +  -n 200 -s record/flows
               + 'duration > 1000' as a filter (to get rid of false
                 positives - flows that have low number of packets (30)
                 and low duration (0.001s) have 30.000pps!
   * flooding from one source to one destination using different port
     numbers.
               + these floods are harder to catch because they appear
                 as different flows.
               + I can see these flows clearly when searching for a
                 specific IP address - which means it's hard to automate.
               + I don't know how to create a filter to generate a top
                 x pps with the flows aggregated by srcip and dstip.
               + I've tried something like:  nfdump  -r
                 /data/nfsen/profiles/live/source/nfcapd.200701222305
                 -n 10 -s record/pps -A srcip,dstip -o extended
               + however, I haven't been able to detect these types of
                 floods - will it work?
   * distributed DOS: flows from numerous sources to the same
     destination, on different ports.
               + I guess it could be implemented like the previous one,
                 but it would be tricky to set the threshold - to
                 separate floods from legitimate traffic


My questions to you are:
1. How do you detect the last 2 types of flooding
2. Have you encountered other types of attack? Can you suggest a better way to detect them?

Thank you kindly,
Adrian Popa


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Nfsen-discuss mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/nfsen-discuss

Reply via email to