% I am not sure you could cover all possibilities of a configuration with 
 % probing for anything more complex than a very simple system.  I think 
 % you need some outside information to guide the probing (e.g. why the 
 % system was set up a certain way, time based dependencies).  When I was 
 % working in the private sector, there were systems that you could never 
 % figure out what happened until you went through an entire yearly cycle 
 % (e.g. month-end, quarter-end, year-end).  You could set up a system to 
 % reproduce the yearly cycle with historical data, but then that is not 
 % just probing a live system.


 While it is true that any system can be tied into arbitrarily complex
 knots, from the outside it generally has to present a narrow set of 
 services. This assumes we are talking about a server providing services.

 But most of the knots seem to be partitionable into problem spaces. 
 IE, the Webserver is different than the accounting package is different
 than the mail package. But there are some expected relationships. 
 IE, the accounting package uses both the webservices and the mail services,
 but this is generally through a known API. Or at least thru a small number
 of countable API's. 

 Since we know something about the structure. And we know the kinds
 of things that are possible, it should be possible to write software that
 find signatures.

 I was looking at honeyd yesterday. It is a honeypot system that creates
 virtual networks including external signatures to trick people with.

 In the same way, it should be possible to look for similar kinds of 
 signatures which guide exploration.


 Let's switch gears... Lets assume we were building an expert system
 to learn how a system admin disects a unix system to understand what it 
 does. And we put a team of people watching you as a system admin
 probing a system. And they ask you, why did you look there? And what
 is the goal of that test? And they build a set of descriptions
 of  tests and explorations....

 is it possible to model that human behavior of exploring a 
 unix system that is unknown?







John Sechrest          .         Helping people use
                        .           computers and the Internet
                          .            more effectively
                             .                      
                                 .       Internet: [EMAIL PROTECTED]
                                      .   
                                              . http://www.peak.org/~sechrest
_______________________________________________
lssconf-discuss mailing list
lssconf-discuss@inf.ed.ac.uk
http://lists.inf.ed.ac.uk/mailman/listinfo/lssconf-discuss

Reply via email to