Re: [sidr] Burstiness of BGP updates

[Brian Dickson <[e-mail]>]

On Wed, Nov 16, 2011 at 8:57 PM, Robert Raszuk <[e-mail]> wrote:
> Hi Russ,
>
> I think the current intention is to secure the network on the basis of
> giving each prefix a badge and just check it at entrance door readers to
> each AS.
>
> If it is allowed in it enters if it is determined by the security back-end
> to be evil it is denied.

That analogy works for packets. It doesn't work for routes.

Routing sucks. Routes attract traffic.

I would instead say, it's like a work order to install a window or door.

That someone with a work order got as far as they did, is either proof
that they are supposed to be there, or a security vulnerability, or
maybe both.

Signed work orders suggest that a window or door was requisitioned. It
doesn't necessarily confirm which wall it should go in - the one to
the next suite, the one to the vault, or the one to the street?

But, chasing analogies to their breaking point isn't necessarily constructive.

Yes, knowing what is intended is necessary if we want to provide
security, as opposed to merely vague assurance.

Given the amount of work expended thus far, taking the extra step
(which may not be large) is both advisable, and something IMNSHO
should be considered in-scope....

Brian