I am writing/reviewing specs for an application level fw, whose sole
purpose in life is to forward smtp mail. Well, from another point of
view, its sole purpose is also to block ALL incoming attempts at access
(<>smtp), naturally...
As I'm fairly new to this fw business, I would like to check following
ideas with you (a bit of free consulting :-)).
The proposed FW system will be configured with as much
hardware/software as needed. Am currently thinking in terms of
(outside)-(r1)-(mynet1)-(fw)-(mynet2)-(r2)-(inside), where everything
between, and including, (r1) and (r2) (routers) is "the firewall
system". The (fw) will be realized with something like fwtk/smap
or a turnkey solution like Gauntlet.
As mentioned, the fw should only forward smtp mail at the application
level. It should NOT, IN ANY WAY, allow any other connections at any
level or for any other application.
I believe that it's possible (even fairly simple:-)?) to create a fw
system as described above (since I only will allow one type of
connection, e.g., I won't need much complicated router filtering
programming).
I also beleive that it's impossible to attack the inside
net through the system, IFF the system isn't otherwise compromised
(the point I'm trying to make is that it shouldn't, as I see it,
be possible to launch an attack _through_ the smtp forwarding
system (no forwarding below level 7)).
Disregarding threats inherent in allowing mail at all (the inside users
are trusted...), what do you think the main threats/riscs are? In what
way can the fw system be compromised?
If anyone has anything to say about these (naive? vague?) ideas, I would
be very grateful.
----
Martin Fredriksson Telephone: +46 31 671000
Systems Integration and Security Direct: +46 31 671036
Ericsson Microwave Systems AB Telefax: +46 31 673771
S-431 84 Molndal, Sweden Internet: eremf @
ere .
ericsson .
se
----
PS. TIS people need not give me too much info, as I will _pay_ Staffan
to help me :-) !
|
|
