Re: [sidr] WGLC draft-sidr-rpki-rtr - take 2?

[Christopher Morrow <[e-mail]>]

On Mon, Apr 4, 2011 at 8:50 AM, Hannes Gredler <[e-mail]> wrote:
> On Mon, Apr 04, 2011 at 08:22:42AM -0400, Danny McPherson wrote:
> |
> | On Apr 4, 2011, at 4:32 AM, Hannes Gredler wrote:
> |
> | >
> | > so my question is: "why do we need to solve the same problem
> | > (= protecting message integrity) 2 times in different ways" ?
> |
> | This new machinery simply introduces object-level integrity functions
> | in the application (i.e., BGP), it does nothing to ameliorate attacks
> | at lower layers - all those substrate attack vectors (e.g., transport
> | connection resets, injection or replay attacks) still exist and
> | require controls as well -- else things might break in even uglier ways
> | at higher layers.
>
> still that does not answer my question: why do we need to solve the problem
> of transport integrity twice (or to play devils advocate:
> shall we encapsulate BGP into SSH up until something better than MD5
> is available ;-))

some folks (not me) suggest that ipsec is the way to go here... (bgp I mean)
I think one point to keep in mind is that tcp-ao has exactly zero
implementations... while SSH implementations abound.

-chris