Up periscope :
> From: horn @
mickey .
jsc .
nasa .
gov
> Date: Wed, 1 Mar 1995 09:23:23 -0600 (CST)
>
> padgett @
tccslr .
dnet .
mmc .
com wrote:
> >So my feeling is that the nodes really do not need any security except as
> >a second line of defense (I like defense in depth - preferably three levels
> >deep) because the problem packets should never reach the nodes if I have
> >done my job properly.
>
> Wow! That's quite a statement. I think that our network perimiter provides
> us the lion's share of our security, but that doesn't mean that I don't make
> node based security a priority.
>
> Do all of you out there really put that much confidence in your firewalls that
> you feel you could comfortably ignore the security of individual nodes? To
> the extent that just about any unix box can be a router and do SLIP/PPP, don't
> I have to worry about a node creating another point of contact from the
> Internet? Or is that kind of awareness not considered node based security?
I'm sure someone will remind us soon that this is a firewalls
discussion, and node-centric security is off-topic. Having 2 cents
burning a hole in my pocket, though...
I think node-based security checks are critical as a defense against
the interior threat. Network protocol analysis will not tell me that
someone has created a bogus root account, or point to illicit setuid
files.
I agree with Padgett on the defense-in-depth. The
hard-outside-and-soft-center analogy is misleading; it implies one
layer is enough. A firewall protects us from the world, and tiger et.
al. protects us from ourselves, and both are justifiably necessary.
down 'scope, rig for silent running...
John Reynolds When action is unprofitable,
Applied Materials gather information.
3320 Scott Blvd. MS 1119 When information is unprofitable,
Santa Clara CA 95054 sleep.
(408) 235-6352 - Ursula K. LeGuin
reynolds @
acetsw .
amat .
com
|
|
