Thanks muchly, all of you.  This'll be a big help.  I'll keep y'all in
the loop for progress on it. 
Kevin Kreamer wrote:
> Elliott Seyler wrote:
>   
>> I'm planning a rather crazy project to make a distributed server, and
>> one of the problems I've come across in my initial planning is
>> preventing people from connecting modified versions of the server.  I
>> want to prevent anything but a legitimate version from being part of the
>> server network, to prevent collusion with the intent to reveal secure
>> information or communication sent or stored within the network. 
>>
>> The trouble is that I can't think of any reliable way to do this.  Any
>> suggestions you may have would be welcome. 
>>     
>
> Unfortunately, the short answer is: you can't.
>
> Longer answer: from the view of your known and trusted server (let's
> call it Rock), all you are doing is sending out requests and getting
> back responses from someone, somewhere.  The issue, though, is that
> whatever Paper (your suspect but actually good server) says can be
> emulated by Scissors (your evil intruder).  An attacker can even go so
> far as having Scissors cut Paper's responses (i.e. run a corrupted proxy
> between two good servers), changing only the responses necessary, and
> leaving everything else the same.
>
> Encryption only gets you so far.  From an encryption standpoint, this
> problem reduces to the problem of secure key exchange.  That in itself
> wouldn't be fatal, but the fact that you want to do secure exchange
> between strangers (and only, somehow, between those particular strangers
>  - killing PKI ideas) is fatal to this idea.
>
> Closed source only gets you so far.  While it is true that in a case
> like this, open source helps an attacker (they change what they want to
>  change, and simply recompile), closed source doesn't prevent an attack.
>  A determined attacker can still reverse engineer the protocol (or run
> the proxy above), and so the only thing closed source buys you is some
> (unknown) amount of time.
>
> Ultimately, your only way to really handle this is to either accept that
>  modified servers will connect (and design accordingly), or personally
> know and trust the people running the servers (i.e. a political solution
> instead of a technical solution).
>
> Kevin
> ___________________
> Nolug mailing list
> nolug@nolug.org
>
>   
___________________
Nolug mailing list
nolug@nolug.org
Received on 05/26/07
This archive was generated by hypermail 2.2.0 : 12/19/08 EST