[dnssec-key-tf] Input for Requirements Requested

On 29 Feb 2008, at 13:56, Daniel Karrenberg wrote:

> If you have any input on requirements, even if it is just in the form
> of "this and that is very important to me", please let me have it  
> quickly.

Just to start the ball rolling....

[1] The TAR should be technology neutral. It should not exclude or  
prevent different flavours of trust anchors to be published.

[2] The TAR should be OS/DNS implementation neutral. Tools and  
documentation should be provided for the common platforms: "here's how  
to transform this tarball of crypto gunk into stuff to plug into your  
name server configuration".

[3] The TAR should somehow verify the keying material it's given  
before publishing or storing it. There should also be a secure channel  
for authenticating the TAR and any data it's publishing.

[4] A process is needed to revoke a trust anchor and notify those who  
may be using the now withdrawn or invalid trust anchor.

[5] Everybody should sign up to T&C's that hold everyone else harmless.

[6] The TAR should be clear what support, if any, is available.

[7] The TAR must make it clear what they keying material is for and  
its political significance: eg "we're not undermining IANA" (or  
perhaps not) or "we make no claims about the appropriateness of the  
stuff in our TAR" (national sovereignty & competition issues).

[8] The TAR should treat all parties equally. Provided they  
demonstrate suitable levels of DNSSEC clue.

[9] There should be regular reviews of the TAR's usefulness: ie clear  
goals for defining "success" or "failure" and some way of establishing  
consensus around these goals.

[10] The TAR must have exit and escrow strategies