RE: Limited vs full blown testing

From: Martin Murray-Brown (Martin.Murray-Brown@derivco.com)
Date: Mon Jun 28 2004 - 03:16:01 EDT


I noticed a couple of people talking about definitions... just so we're
all on the same page, perhaps we should agree on the following:

1) A 'DoS' attack is a Denial Of Service attack (and not the operating
system ;) ). In other words, it's any attack that results in a denial of
service... as Alan said, stealing your keyboard and mouse would be quite
an effective DoS attack, especially if you don't have a spare :)

2) 'DDoS' is 'Distributed Denial of Service'... an attack where multiple
clients (often viral zombies) spam a particular node in some way,
preventing that node from receiving valid requests. Assorted flavours
include reflected (where the node is not spammed directly, but rather
hit with response packets from spoofed IP's on the original packets...
nasty).

Therefore, a DDoS is a DoS, but a DoS isn't necessarily a DDoS. Groovy
*snaps fingers*.

In terms of threat... while possibly I missed the original point of this
which restricted us to penetration tests, I still believe that any
remotely complete test requires some form of DoS testing. In terms of
damage to an online company's cash flow, DoS's can be devastating... I
recommend that any proposed test that doesn't include denial of service
testing ensures that the client is fully aware of the ramifications. You
don't want big clients coming back and blaming you for not telling them
about it when some Skiddie with a few hundred zombies is costing them a
million a day...
(I know I'm harping on about it and repeating what others like Alan have
said in different words... but DoS's are becoming more common, and are
being used in blackmails (check recent reports in the online betting
industry).

-----Original Message-----
From: Alan Davies

>I'm trying to understand the significance of DDOS testing and
importance.
>Thing is, if you can spew packets fast enough, or make enough
connections
>to consume the resources involved, you can take a site/serice down for
at
>least the duration of the attack, even pipes as large as those of
>akami<sp?> were proven to be susceptable in recent days. It's a given
>vector of attack that we live with, a risk level we hope to avoid.
But,
>not something that gives away the insides of the network to thugs and
>theives. No root shell and all that, which constitute a real threat,
at
>least in my mind. Perhaps I'm missing something that has come up in
>recent years that redefines DDOS as something that is preventable and a
>potential for something other then a blip, however long lasting the
>attack, in service?

Ron - I think the difference here is DoS vs. DDoS. The latter is just
throwing packets at a target to fill all available bandwidth and I can't
see
a lot of point in that during a pen test (in that it's not actually
compromising anything).
 
However a DoS can be anything that denies service - if I walk up to your
desktop and steal your keyboard and mouse, I've DoS'd you by stopping
you
working ;) Seriously though - run Nessus with dangerous plugins on and
you
will likely DoS many parts of the clients network .. and not by
overwhelming
with packets. You may find that some routers/switches have been killed
until a full power cycle is done and that some systems (especially
older)
have completely and irrecoverably locked up. It could even end up
causing
data loss.
 
The fact of the matter is, if there are systems that can be knocked down
like this by an exploit, then you would really want to know about it and
try
to prevent it. At the same time, if the client is aware of this and
doesn't
want to take the risk ... well they are the ones paying you and all you
can
do is tell them!
 
 
 
P.S. One final reminder of how a DoS can be used in a penetration ....
think of good old Kevin Mitnick! Without DoS he wouldn't have been able
to
break in the way he did.
 
 
 
Best regards,
 
 
Alan Davies.



This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:53:57 EDT