FW: Signature on logs/eMail

From: Everhart, Glenn (FUSA) (GlennEverhartat_private)
Date: Wed Jul 25 2001 - 05:43:23 PDT

  • Next message: mat_private: "README.MODERATOR"

    Re: log file protection.
    
    A WORM can be used like this. To use with normal filestructures you need to
    segregate the filestructure stuff (index files, directories and the like)
    and allow data to be written to the disk, or have a filestructure that
    just looks for the highest version of a directory or index block. I wrote
    a software WORM and published it (in src) in the mid 90s for VMS; the driver
    checked that any disk block it wrote above a "fence" block number was
    being written for the first time and would junk the write and claim success
    if not (so someone messing with it would not immediately realize that
    the attempt failed). The container was encrypted to defend the abstraction
    as I
    didn't want someone to mess with the container file directly. Someone could
    of course corrupt the contents, but it would be difficult to alter them
    tracelessly. Note too that the mapping of container file to logical disk
    blocks can be whatever I want. Not totally bulletproof, so a real WORM
    device is far better, but good enough to resist many attacks, in the sense
    of making tampering more likely to be detectable than in cases where a
    logfile is just written somewhere. The reason for doing this kind of thing
    was to improve the integrity of logfiles where it was infeasible for
    whatever
    reason to alter the software that wrote them.
    
    I should note that most sequential writes in fact buffer a device block
    before
    writing to a device, and that these techniques are useless against attacks
    that alter the memory buffers before they can be written. Clearly, more
    verbose logs minimize this issue.
    
    A number of companies have filesystems for WORM devices. The ones that write
    data to WORM periodically use this "normally use the highest version" scheme
    but can generally be told to use earlier indices to in effect undo changes.
    Glenn Everhart
    
    
    -----Original Message-----
    From: rferrellat_private [mailto:rferrellat_private]
    Sent: Tuesday, July 24, 2001 9:01 PM
    To: forensicsat_private
    Subject: Re: Signature on logs/eMail
    
    
    Hi folks,
    
    One of the solutions to the log authentication problem I've been pondering
    lately is the write once/read many network logging device.  This could be
    essentially a CD-ROM drive (or the equivalent technology) designed to write
    a copy of the log in real time, using a process that makes subsequent 
    overwrites
    physically impossible (or as near to impossible as one can get in this 
    business),
    at least without obvious signs that an overwrite has been attempted.  We
    know,
    for example, that each write pass over magnetic media leaves a permanent
    trace in the media substrate, so a detector specially designed to check for
    multiple writes could be employed as a verification check before and after
    the media was loaded into the drive.  It could even burn a permanent record
    of the initial verification run, protected by a checksum (for example) in
    the
    header of the disk itself.
    	This is all highly speculative, of course, although some aspects of
    this technology do already exist, and it doesn't address the issue of 
    verifying
    currently existing logs.  It would be expensive to develop and, at least 
    initially,
    to deploy, but it might be a viable long term means of coming to terms with
    the electronic record verification problem.  In thinking about the problem, 
    I was
    reminded of my days in corporate security, when alarms and other 
    electronically
    monitored security incidents were recorded in real time on a dot matrix 
    continuous
    feed stack, attempted alterations to which were usually quite easy to 
    spot.  Courts
    seemed to have little trouble accepting those logs as genuine.
    
    Just a little postulating from a speculative fiction writer...
    
    ;-)
    
    Cheers,
    
    RGF
    
    Robert G. Ferrell, CISSP
    
    
    -----------------------------------------------------------------
    This list is provided by the SecurityFocus ARIS analyzer service.
    For more information on this free incident handling, management 
    and tracking system please see: http://aris.securityfocus.com
    
    -----------------------------------------------------------------
    This list is provided by the SecurityFocus ARIS analyzer service.
    For more information on this free incident handling, management 
    and tracking system please see: http://aris.securityfocus.com
    



    This archive was generated by hypermail 2b30 : Wed Jul 25 2001 - 09:45:58 PDT