Firewall Wizards mailing list archives

Sidebar on a historical security model


From: David Collier-Brown <davecb () canada sun com>
Date: Fri, 30 Oct 1998 14:19:34 -0500

Paul D. Robertson wrote:

      Borrowing from the ``medieval city'' metaphor, the
      machine would serve as the gate, the public market
      inside the gate, and the gate in the inner marketplace
      wall. You still have to hire some spear-carriers
      to stand at the gate and catch theives, though.

Certainly, but you'd rather the gatekeepers had strong spears and a good
portcullis than a hole in the wall and a stick.

        Definitely: we're  re-inventing the authentication,
        authorization and transitive trust problems that
        we re-solved in the middle ages.  (Re-solved 
        because the Greeks & Romans were there first and
        got the T-shirt).

        [Let me blather for a moment and I promise to get back
         to the point]

        The problem was hackers with swords.  The solution
        was heavyweight: walled cities.

        The solution produced a larger problems: armies with
        ladders.  The solution to this was castles (ok, it 
        was actually in the opposite order, castles spawned
        some of the cities). If an army got over the wall,
        everyone retreated to the castle.

        Individual families built houses with castle-like features,
        gates and walls.  The front and back rooms became stores
        and factories... and the front door was the gate.

        The guy in the castle had a staff and soldiers who wore 
        his badge, so he effectively authenticated them. Rich
        families did exactly the same thing.   The guys on the
        front gate honored the lord's badge and often the 
        family badges even after dark and when the gate was closed.
        They tended to check more carefully then, of course.

        A peasant from the country or a traveling salesman
        could wander in the gate, be passed by the guards
        and be told where to rent a room, and wander into
        anyone's store, but not jump over the counter.

        [Return to modern times...]

        This is actually a very rich feature-set for an
        authentication/authorization/transitive-trust
        system...  better that we have with firewalls
        and logs of passers-by.
        

VPNs, telecommuters, extranets, and all the other things we're being
avalanched with necessitate per-service security policies instead of
per-network ones, I see this as a probable avenue into that realm.
There's enough code in the system now to do good things, and enough left
to do to shape what can be done.

        I think it's one of the primitives we will end up
        reinventing, in the move from walled cities to the
        very relaxed but still effective models we have today.
        
        I claim the current model is far better, and much lighter
        weight.  I don't **want** to be a wizard: I want things to
        be really easy.  And reducing the cost hurdle and raising the
        allowable-stupidity hurdle is a step on the way.

        
--dave 
[ps: I think I could write a good analysis of this as
     a series of historical solutions to a safety-critical 
     problem, giving us a family of models to argue about.
     I'm very unlikely to do so unless my employers' security
     folks decide they need a philosopher.  It's unusual
     enough that the ABI folks decided they needed one!]
-- 
David Collier-Brown,  | Always do right. This will gratify some people
185 Ellerslie Ave.,   | and astonish the rest.        -- Mark Twain
Willowdale, Ontario   | davecb () hobbes ss org, canada.sun.com
N2M 1Y3. 416-223-8968 | http://java.science.yorku.ca/~davecb



Current thread: