Re: Re: RR checks to avoid DoS attacks

Bill,

I am not saying that we ignore realistic DoS attacks, but I think this 
particular scenario we are examining is not a practical avenue for an 
attacker.  I feel strongly that the cost of the attack is sufficiently 
large for the attacker, and comfortable with not doing an RR.

If I am hand-waving in characterizing this particular attack you 
articulated as unrealistic, please make that case.  Thanks.

Also, please see inline:

Bill Sommerfeld wrote:

>>If the said DoS attacks are the only reason for RR checks in IKE based 
>>signaling of mobility, I think that part of the protocol should be optional.
>>    
>>
>
>The widespread deployment of nodes which can be exploited into
>performing DoS attacks on a third party victims is the single biggest
>security problem facing the current Internet.  
>
>  
>
I would say 'one of the problems,' but agreed in general.  It is not a 
problem however when the potential attacker is cryptographically 
authenticated to the entity that facilitates the attack and possibly 
incurs a higher cost ((the IPsec GW has to encrypt and encapsulate the 
packets, whereas the victim just drops them) than the targeted victim.

regards,
Lakshminath

>I would be most upset if we ended up deploying another such set of
>nodes as a result of this.
>
>We absolutely must examine each potential DoS attack; we cannot, as
>was once common, wave our hands and say that dealing with DoS attacks
>is impossible.
>
>						- Bill
>						
>
>  
>