[RRE]"code red" worm

From: Phil Agre (pagreat_private)
Date: Sun Aug 05 2001 - 14:34:32 PDT

  • Next message: Phil Agre: "[RRE]pointers"

    A new worm, apparently completely different from Code Red but taking
    advantage of the same hole in Microsoft's server software, began
    spreading yesterday and is now being frantically analyzed by security
    people.
    
    The initial report may have been this one (read the subsequent discussion
    as well):
    
    Code Red II Worm Analysis
    http://www.dslreports.com/forum/remark,1226089;root=security,1;mode=flat
    
    Some draft analyses are here:
    
    SecurityFocus Code Red II Information Headquarters
    http://aris.securityfocus.com/alerts/codered2/
    
    CodeRedII - New Non-Variant CodeRed Worm - Analysis
    http://www.securityfocus.com/archive/75/201877
    
    CodeRed II ARIS Incident Analysis
    http://www.securityfocus.com/archive/75/201878
    
    Here is the URL for current discussion on the Incidents mailing list:
    
    http://www.securityfocus.com/frames/?content=/templates/archive.pike%3Flist%3D75
    
    Here is a broader analysis inspired by the first Code Red worm:
    
    How to Anonymously Get Root Access on a Quarter Million Machines Overnight
    http://braddock.com/cr2.html
    
    Here is another site for analysis of the first worm:
    
    http://aris.securityfocus.com/alerts/codered/
    
    I'd also like to take a moment to vent about the so-called journalists
    who have been clucking about how overblown the Code Red hype was.
    Yes, as I pointed out here, there was some hype, in the form of reports
    that made it sound like regular users who aren't running servers needed
    to download the IIS patch.  But that hype was perfectly understandable
    given the very real need to distribute an alert combined with the
    extremely technical nature of the problem.  And yes, the worm was
    "never dangerous" in the sense that, by pure luck, it had a bug that
    prevented it from carrying out its planned destruction.  But these
    faux-knowing "it's all hype" types of reports are grossly irresponsible
    nonetheless, for several reasons:
    
    (1) Anything that invades a quarter-million servers is a serious matter. 
        Hello?  Do we need any more reason to be raising alarms about the
        state of our information infrastructure?  One particularly idiotic
        columnist even said, who cares if some eBay bids don't go through?
        I don't have words strong enough to denounce this foolishness.
    
    (2) The fact that the worm had a bug was not remotely obvious at first.
        It didn't become clear until some very smart people worked a lot
        of hard hours to capture a copy of it, disassemble it, study it,
        and run experiments on it.  Have a good look at the URL's above
        if you want to get a clue of the kind of work that goes into this.
    
    (3) And even then the security people couldn't really be sure how
        the worm would interact with the full range of real systems out
        there in the world, with a quarter-million copies of the thing
        all probing machines at random to see what they could find.
    
    (4) The worm *could* have been extremely destructive through simple
        changes to its code.  A distributed denial-of-service attack
        from a quarter-million servers -- that is, an attack in which a
        quarter-million of the most powerful, highest-bandwidth computers
        on the whole Internet send complete junk onto the network at their
        maximum capacity at the same time -- would bring large parts of
        the Internet to a halt.  These attacks are nearly impossible to
        defend against, and the sites to which the junk is directed would
        be off-line for the duration.  Nobody knows for certain how the
        network itself would react to that kind of load, but we do know
        that traffic was badly disrupted by the train fire in Baltimore
        last month that cut a single link.
    
    (5) The most destructive features of the worm were actually backward by
        the standards of current worm engineering.  The worm was programmed
        to attack a single site, but other worms are programmed to lie
        dormant until they receive orders from headquarters.  Much more
        sophisticated schemes are easy to imagine.
    
    (6) The vulnerability that the first Code Red worm exploited is still
        out there.  A new worm could come along and infect the exact same
        quarter-million servers within about twelve hours at any time.
        That is what is appears to be happening now.  The worm-writers
        read the trade press and the online security sites, and they know
        very well what went wrong with the first worm and how to fix it.
    
    (7) Both the first and second waves of that worm infected something on
        the general order of a quarter-million servers, even though between
        the two waves the vendors, government agencies, and media flooded
        the airwaves with publicity urging IIS server administrators to
        download the patch.  Yes, I realized that some machines did get
        patched, and that the second wave of the worm may have been able
        to attack machines that the first version could not.  Nonetheless,
        we are talking about the general order of a quarter of a million
        servers.  Short of tracking all quarter-million of them down
        individually and yelling at them, nobody has any other way of
        compelling these server administrators to download the necessary
        patches.
    
    (8) Server administrators, furthermore, do not have enough incentives
        to load the necessary patches.  These so-called zombie worms do not
        inflict their main damage on the servers they invade, but on the
        third-party sites that they attack.  This is what economists call
        an externality.  If server administrators had effective liability
        for the damage that their machines are used to attack then they'd
        have the necessary incentives, at least under idealized economic
        conditions.
    
    (9) The Code Red worm was not (i.e., is not, because it's still alive)
        a fluke.  It exploits a type of security vulnerability -- a buffer
        overflow -- that results from grossly shoddy engineering, but that
        has nonetheless been seen many times, in many products from many
        vendors.  The worm is, in other words, evidence of a systemic
        problem, and one that is getting steadily worse as black hats
        get more and more bored with not destroying the information
        infrastructure of the whole world.
    
    (10) The markets for many of the most security-sensitive categories of
        software are highly concentrated.  In the present case the vendor,
        Microsoft, has been found to be a monopolist by a federal appeals
        court.  Because of its market power, it experiences little market
        pressure to reform its shoddy engineering practices.  Instead, it
        invests in dime-a-dozen propagandists who fashion misleading sound
        bites that dissociate responsibility from their own firm, either
        denying that the security vulnerabilities in their products are
        actually vulnerabilities, or that the users are at fault, or that
        it's about computers in general and not their own products in
        particular, or that the threat from worms is all hype.
    
    Am I predicting that the Internet will collapse?  No, what I am doing
    is yelling at the bogus reporters (not the legitimate ones, who are
    great, just the losers) to get away from the simplistic dichotomy
    of "it's the end of the world" versus "it's all hype".  This worm
    has already mobilized the security community, who are hard at work
    preparing alerts, remedies, patches, counterattacks, and everything
    else they can think of.  They will probably succeed in preventing
    the Internet from shutting down, and more power to them.  When they
    do, let's not talk about what hype it all was.  Instead, let's talk
    about what heroes the white-hat security people are, what long hours
    they put it, and what a criminal shame it is that such heroic efforts
    were necessary to prevent critical infrastructure from collapsing.
    
    Phil
    



    This archive was generated by hypermail 2b30 : Sun Aug 05 2001 - 15:38:47 PDT