Re: [RPSEC] BGP Security Requirements v08

At 5:44 PM -0400 7/10/07, Sandy Murphy wrote:
>  > My comment was:  One might interpret this to mean that any extended
>>  form of UDATE message MUST be transmitted through ASes that are not
>>  capable of processing it. This is imposing an unnecessary burden on
>>  solutions, since an alternative is to just mot send such messages
>>  through ASes than can't make use of them.
>
>This is a question about use of security information during initial
>deployment.  If there are islands of deployment that must communicate
>through seas of non-deployment, is it useful for the deployer to
>learn that there is a protected route island out there?  When the
>information came to them over an unprotected sea?
>
>I'd say that there is a potential for benefit.  Perhaps the unprotected
>sea in between is known by other out-of-band means to be reliable,
>even if clueless about the new protection mechanisms.  There might
>be operational procedures that protect the intermediate sea that
>are adequate from the point of view of the recipient on the island.
>
>So I'd say that passing the info along is potentially beneficial,
>to some, and should be passed.  The cost to the clueless routers
>is storage and bandwidth - processing cost should be minimal.
>
>--Sandy

Sandy,

You make a good point about the potential benefits of being able to 
pass updated UPDATEs through ASes who are not able to make use of 
them. However, the benefit is only a potential one, and the flip side 
is that the intermediate ASes might not be able to store the data 
from the UPDATEs, which would thus disallow the solution approach. 
If, as Tony suggested, we want to avoid imposing undue constraints on 
solutions, then we ought not impose this specific one.

Steve