Re: [dnsext] Historical root keys: The Large Router Vendor Speaks

[Paul Wouters <[e-mail]>]

On Fri, 28 Jan 2011, Paul Wouters wrote:

> On Fri, 28 Jan 2011, Phillip Hallam-Baker wrote:
>
>> Beyond that, I do not like the idea of a chain of historical roots. It 
>> means that compromise of any root will poison the ones downstream. There 
>> are five
>> potential points of compromise in the following chain:
>> 
>> A -> B, B->C, C->D, D->E
>> 
>> This approach only ever has two at any given time:
>> 
>> A->B, A->C, A->D, A->E
>
> But that requires the key move somewhere trusted so it can make a NEW
> signature for the A->C transition, after A has been retired for B. Or
> rather, when you deploy E, you have 5 root keys to protect instead of

Oops, as Phillip said, it is only two. I misunderstood. But it still requires
two private root keys to be available, instead of one.

Paul