Re: Secure host newbie - fun - humm

[Barry Fitzgerald <bkfsec () sdf lonestar org>]

Ranjeet Shetye wrote:

> Please dont tell me how much I have or have not worked with security.
>
>  
Please don't make statements that leave only the conclusion that you 
haven't worked in security for long, if at all.

> For security to work correctly, the user should have black and white
> choices. For the user to make a concious choice between secure and
> insecure mode, the person deploying it should offer black and white
> choices (e.g. hotmail login). In order that the people deploying the
> solution can make concious black and white choices, people who design
> the solutions have to MAKE and OFFER black and white choices. 
I agree with your point here.  It's too bad you didn't say that in the 
first place.

> This is
> where I come in. I do network protocols design, security solutions, and
> kernel work for a living.
>  
Congratulations.


> Now if people in my position were to not view security as a black or
> white issue, the deployers would end up with a greyish "solution", and
> the users would end up with a smudgy hazy grey that would be wrongly
> passed around as a secure solution.
>
>  
I never once advocated that you give people a "hazy grey"  solution.  
Not once did I ever say that. 

However, as an implementer, it is your obligation to understand the 
nature of insecurity.  The nature of insecurity *IS NOT* black and 
white.  If you don't understand why that is, then you are doing a 
disservice to your customers.  What you know to be true and what you 
show your customers are not the same thing.  You engineer your solutions 
for your customers with the knowledge that you have. 

Looking at security as black or white, you invariably have to box 
yourself into the belief that all you know about security procedures is 
all there will ever be, not looking for the unorthodox things along the 
way.  If you are looking for unorthodox methods of compromise, then 
you're admitting that security is not black and white.

Which is it? 

> You are focussed on the fact that the "real world" is not black and
> white. 
I'm focused on the fact that security is not black and white. 

> In my job, I HAVE to view security as black and white, and design
> stuff accordingly, and offer choices accordingly, because the farther up
> this design chain that someone compromises the concept of "either
> completely secure or else its insecure", the more compromised the final
> deployed solution that the users actually use. 
Yes, in your position.  But, again, saying that and turning around and 
saying "security is black and white" are two entirely different things.  
If you can't speak outside of your experience, then you simply 
shouldn't.  You had a good point initially, it's too bad you had to take 
my disagreement and turn it into a "yes/no/yes/no" argument about the 
nature of security.


> e.g. I can design the
> most secure system in the world, but that's meaningless if the users put
> their password on a sticky on their monitor. On the other hand, if I
> allow the use of telnet to backup sensitive data like passwords, then
> the I've just compromised even the most savvy user, and I force them to
> work outside the system to achieve their security goals. Therefore,
> should telnet be allowed ? NO - its a simple, black and white decision.
> You have a vendor who forces telnet onto you ? then you have a vendor
> who doesn't care about the security of your digital assets and I suggest
> that you vote with your money.
>
>  
Again, you're boxing yourself into your own experience again.  When I 
said there were valid uses for telnet, I didn't mean for people to get 
boxed into using telnet.  I agree completely and that's why I labelled 
telnet as insecure.  However, if you have assets that are on an enclosed 
LAN with no internet access or that aren't that valuable to you, then a 
lightweight protocol like telnet might be a better engineering choice 
than a heavier protocol like SSH.

That's no black and white there - it's gray... again.  Encryption 
doesn't inherently mean security and likewise, not being encrypted 
doesn't mean inherently insecure.  It depends on your context and your 
resources. 


> As I said before, if you have run systems where there is no known
> weakness - that's secure. If you run systems where there is a known
> exploit - that's insecure. 
No - it's "believed to be secure"... there's a major difference. 

If you advertise and guarantee security in the fashion you have above, 
prepare to be sued by your customers.

> That fact that you "know" that all boxes in
> the world are insecure is not the "know" that I am talking about. 
But, it is a form of knowledge that matters.  Like I said, you can stick 
your head in the sand all you want -- that's just being ignorant.

You advocated people shutting off production boxes because of the 
*possibility* of them being attacked.  You should be prepared to back 
that statement up in all cases... or concede that it's not the correct 
answer.

> I think we should just agree to disagree.
>
>
>  

But, then I'd be doing you a disservice.  :)

               -Barry



---------------------------------------------------------------------------
Ethical Hacking at the InfoSec Institute. Mention this ad and get $545 off 
any course! All of our class sizes are guaranteed to be 10 students or less 
to facilitate one-on-one interaction with one of our expert instructors. 
Attend a course taught by an expert instructor with years of in-the-field 
pen testing experience in our state of the art hacking lab. Master the skills 
of an Ethical Hacker to better assess the security of your organization. 
Visit us at: 
http://www.infosecinstitute.com/courses/ethical_hacking_training.html
----------------------------------------------------------------------------