Re: SANS & Ranum on DoS Trojans for Solaris

From: Marcus J. Ranum (mjrat_private)
Date: Wed Jan 05 2000 - 08:08:09 PST

  • Next message: arkat_private: "Re: Sizing a firewall"

    Vin McLellan wrote:
    >        Want to tell us about this tool Dave Dittrich and you developed to
    >scan network hosts for Solaris machines infected with trojans which install
    >clients distributed Denial of Service attacks:  trinoo, TFN, TFN2000, or
    >stacheldraht?  
    
    
    Dave gets the credit, all I did was code cleanup, portabilizing,
    and some optimization. But since you've asked, I'll explain how
    it works.
    
    There's a whole generation of denial of service tools that have
    been released lately, that do distributed DOS attacks. TFN, Trinoo,
    Stacheldraht, etc, all operate by having a master controller program
    that activates agents residing on multiple compromised machines.
    The agents know how to launch various types of denial of service
    attacks. Agent/master communications are encrypted. The weakness
    in the tools is that the agents and masters have to speak somehow,
    and there's a "ping/alive" capability whereby the master can identify
    active agents to use in launching an attack.
    
    Dave's tool works by emulating the master's pinging, to get any
    live agents to answerm - essentially giving themselves away. You
    give it a class B network (with various masking options so you can
    select down to class C or individual machines if you want) and it
    just searches each host for an agent, by emulating a master controller.
    
    The bad guys will doubtless respond by changing the default
    encryption keys, etc, which will make these kind of tools less
    effective. The good news, for now, is that most script kiddies
    aren't doing that and denial of service attacks are the kind of
    attacks that only appeal to script kiddies in the first place.
    Even so, we're fortunate that the hackers that build these kind
    of tools don't really understand computer security, or they'd
    realize that the systems they build are vulnerable to traffic
    analysis. In the large, you find that, even encrypted under
    different keys, the traffic between a master controller and
    agents will have a very distinctive fan-out and back/forth
    pattern. Of course, to detect that kind of thing, you need
    broad-based network analysis tools. :) I've known that for a
    while. Hence NFR. :) There's a set of N-code filters for detecting
    Trinoo/TFN on our web site at http://www.nfr.net/updates/ if
    you want to see how they work.
    
    mjr.
    --
    Marcus J. Ranum, CEO, Network Flight Recorder, Inc.
    work - http://www.nfr.net
    home - http://www.clark.net/pub/mjr
    



    This archive was generated by hypermail 2b30 : Fri Apr 13 2001 - 13:56:44 PDT