From: nick at virus-l.demon.co.uk (Nick FitzGerald)
Subject: DCOM RPC exploit  (dcom.c)

"Admin GSecur" <admin@...ernmentsecurity.org> wrote:

> I completely agree, unfortunately this is a constant problem in any
> enterprise size network.  So many times it only takes a less experienced
> network admin to bring a network to it's knees.

True, but even that can be mitigated somewhat -- of course, it costs 
money and reduces the "flexibility" that computers are "expected" to 
provide so few apparently even consider it...

> Personally I have found an aggressive and continuous network auditing
> policy by the corporation can help negate any difference in security
> levels among the large number of servers and devices on their network.
> But so many times the corporation doesn't want to spend the capital on
> the manpower needed. (Even if it is for a single additional body)

Managed switches, strict MAC-to-port mapping, etc, etc.  Yep -- costs 
money to set up (more expensive hardware) and a little more _initial_ 
admin time, but it pays back in spades.  Worried about stray (or any!) 
WLAN APs turning up inside your network?  Well, if their Ethernet MAC 
can't talk to any of the LAN, it can't cause any more damage than to 
the person or persons conniving _against policy_ and thus liable to 
being reprimanded, lose bonuses, be fired etc _when_ caught (note, not 
"if" because your auditing will tell you an unknown device was plugged 
into a specific outlet in a specific office/cubicle/etc -- an act you 
can make a policy violation in itself!).  Worried about contractors and 
consultants who drag their laptop all round other people's network then 
plugs into yours?  Don't give them service until their machine has been 
through a specified vetting process.  Costs a bit of time at their 
outrageous charging rates, but is your network security really _not_ 
worth that effort?  Worried about that unauthorized ("black ops") 
server the guys in accounting have slipped in the backdoor?  You need 
not.  It can be done -- it simply takes a bit of structure and 
management which is what I've always thought being a system 
administrator was all about.

> Another common problem is the random "orphaned" server that was
> installed with out the NOC ever knowing.  So many times this single
> neglected machine can cause the downfall of even the tightest network.

True, but not in a well-designed and administered environment.  The 
idea that any one should be able to connect any device to the LAN was 
maybe just acceptable in certain environments _BEFORE_ those LANs were 
attached to the completely inappropriate to task "modern Internet".  
Such a network "structure" is utterly unacceptable in today's world of 
massively increased connectivity, massively increased staff 
accessibility to computers (both at work and elsewhere), massively 
increased mobility of the enterprise's computers and the fabulous 
arrogance shown by the preferred supplier of the OS running most such 
computers in ignoring entirely for the about 5-6 years the complete 
change in the security and system integrity landscape the massive 
acceptance of "the modern Internet" should have entailed.

> So are the woes of an enterprise admin.

...if s/he is under-resourced.  It need not be the way you describe at  
all.  It can be done _much_ better with a relatively small initial 
commitment to "doing it properly".  Of course, most enterprise systems 
start off with the "is there a cheaper option" attitude.  In a 
"manufacturing" type operation, such a question tends to have a 
(relatively) low opportunity cost should it turn out that the cheap 
route leads to too many failures (you switch to the slightly more 
expensive but better quality inputs, suck up the cost of recalling or 
otherwise replacing the already shipped stock and perhaps push up the 
price a bit).  However, if what you scrimped on was "infrastructure" 
fixing it nearly always entails pulling out great gobs of now largely 
useless and valueless plant and replacing it with more expensive but 
better fit equipment.  Thus taking the cheap route on infrastructure 
such as networking equipment, operating systems, network security and 
so on becomes a hugely expensive thing to "fix properly".  Thus we tend 
to see band-aid fixes such as firewalls, IDSes, antivirus software and 
all manner of other things that should be largely unnecessary were the 
system they are going into "properly designed" from the start.  (This 
is not say that those things might not have _some_ value in a properly 
designed and maintained system, but they certainly should not be the 
security "front line" in enterprise systems.  Of course, for the SOHO 
market, such band-aids may be the only viable solutions given there is 
not (enough) money to hire the best trained and equipped professional 
sysadmin staff nor can the initial setup overhead of "doing it 
properly" be justified against the size of the userbase and/or the 
value of the "operation".)

Of course, convincing a bean-counter of the value of taking a longer-
term view of such issues is really difficult and almost exclusively you 
will only ever find such principles applied in practice at _extremely_ 
sensitive installations and at large corporations that have been "hit" 
very severely because they got it wrong the first time.  After seeing 
the lack of value of scrimping on critical infrastructure there is a 
tendency for upper management backing for "doing it right" the second 
time around.  I guess that this is almost exclusively how it is means 
the "it won't happen to us" attitude is alive and well in the halls of 
corporate governance...


-- 
Nick FitzGerald
Computer Virus Consulting Ltd.
Ph/FAX: +64 3 3529854

Powered by blists - more mailing lists