Re: [DNSOP] draft-ietf-dnsop-dnssec-trust-history - discussion

[Joe Abley <[e-mail]>]

On 2010-09-14, at 13:55, Tony Finch wrote:

> On Tue, 14 Sep 2010, Joe Abley wrote:
>> 
>> Doesn't trust-history impose a requirement high standards of operational
>> security for key materials which have long since fallen out of
>> production, and hence extend the possible window for a key compromise
>> long after the key has stopped being used? From an operational
>> perspective this worries me.
> 
> I haven't checked the draft, but it should be possible to throw away a
> private key after it has signed its successor and been decommissioned.

Right. I'm concerned about the scenario where

 - KSK goes out of production
 - KSK is supposed to be destroyed following some suitable emergency roll-back window
 - KSK is not actually destroyed due to operational error
 - KSK is compromised by someone (hard disk found in dumpster)
 - KSK is used to create an apparently-legitimate branch of the trust-history chain, legitimising a bogus key

In normal operation this would not be a concern because the retired KSK is of no practical use. By inventing a practical use for the retired KSK, the requirements for secure handling are extended until the end of time.

I agree that in an ideal world the KSK would be securely destroyed. I don't often feel like I live in that world, however, hence my worry.


Joe