At 08:37 03-07-2007, Hallam-Baker, Phillip wrote:
Arguments about false senses of security are usually wrong. We are
adapted for an environment where sabre toothed tigers are running
around at night. So our tolerance for insecurity is much higher than
you might think. A sense of security is created by familiarity, not
by analysis.
NAT has coined an idiom that you are secure by virtue of using
it. That's not stopping the tigers.
The point here is not to shut off connectivity, but to shut it off
by default. If the user knows they want to run an HTTP server they
can set it up accordingly. The nice thing about NAT boxes is that a
user who knows nothing about their network can plug one in and the
default state is to deny inbound connections.
Shutting off inbound connectivity comes five years too
late. Nowadays, the treat is also from the inside. NAT makes it
more difficult to identify and take measures against the offending host.
What that means is that the MYSQL server or the embedded SQL Server
in their application they didn't even know they had is shut off from
external access. I was running an unpatched version of SQL server
when slammer hit without ill effect. I have boxes that have not been
booted for several years. I'll bet that much on the code on those
boxes is vulnerable too. I don't patch systems I am not currently using.
That $50 box can be set by default to deny incoming connections so
that the user than knows nothing about their network can plug it
in. Outgoing connections could be restricted as well so as to deny
these unpatched systems external access by default. Currently, this
is not viable due to IP addressing constraints.
Sure you can do much better if you 1) know what you are doing and 2)
are prepared to put in the necessary time. Most people don't meet
either condition.
I agree. Any solution for the type of user you mentioned is only
effective if it is easy for them.
Regards,
-sm
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf