Re: [sidr] BGPSec scaling (was RE: beacons and bgpsec)

[Rob Shakir <[e-mail]>]

On 7 Sep 2011, at 14:12, Jakob Heitz wrote:

> While a router that performs BGPSEC may not be more expensive in 5 years than one that does not today, that is not relevant. A router that performs BGPSEC in 5 years will most definitely cost more to produce as well as cost more to run than a router that does not perform BGPSEC in 5 years.
> 
> So, a question for you Rob. Will your customers pay the premium for BGP security?

Hi Jakob,

This is of course an interesting question - which comes down to the question of whether the threats that are being addressed by bgpsec are common-place. I definitely have customers that would pay a premium to mitigate this as a DoS vector, or malicious interception mechanism, but equally, have customers who would not, based on their current experience.

From what I have seen of the demand for origin validation at the current time, I would say that my personal opinion (and no dataset to support this, sorry) is that any willingness to pay a premium will grow relatively slowly. As such, this makes the point about trying to ensure that we have a deployable protocol that attempts to represent the smallest step change it can in terms of computational requirements more important to me - since this will mean that it is easier to begin deploying, and meeting the demand.

Kind regards,
r.