Re: additional attack for multi6 threat draft?

[Masataka Ohta <mohta@necom830.hpcl.titech.ac.jp>]

marcelo;

> > > Why?
> >
> > Why?
> >
> > Can you name some protocol that does not do this?

> Take mip and remove the rr check.
> You are supposedly talking to the HoA but you are sending packets to the
> CoA, so you don't have a RR check of the address that you are supposed to be
> talking to i.e. the HoA

You are talking about triangle elimination, which, as I mentioned, is
a cause of new form of redirection attack.

However, it is purely a mobility issue having nothing to do with
multihoming.

A CN and a MN can use all the complex security mechanism to locate
and identify each other. However, it does not prevent MN redirection
of traffic of CN to a DOS victim. The victim can be protected from
DoS, if 8+8 is used, though the network around the victim still suffer.

> Again, RR is fine to verify locators but it is not so great to verify
> identifiers.

RR check involves cookie exchange, which is to verify the identity.

> In multi6 where we may need redirection, you cannot assume that RR will be
> available to verify identities

Cookie is our friend.

							Masataka Ohta