Re: [babel] I-D Action: draft-ietf-babel-information-model-02.txt

[David Schinazi <[e-mail]>]

> On Apr 10, 2018, at 21:05, Markus Stenberg <[e-mail]> wrote:
> 
> On 11 Apr 2018, at 2.04, David Schinazi <[e-mail]> wrote:
>> 2) Regarding security, should it be possible to pull the private key from a router? Is that common practice?
> 
> No. I think what is written in the spec now is good.
> 
> In general only public keys should ever go anywhere from a device. Any scheme that involves private key distribution is quite a bit less secure.
> 
> That’s why e.g. certificate (= glorified public key) enrollment practises involve device making requests with the public key, and getting signed one back in return. Similarly, why HSMs and such are all the rage (keep your private key on a box you don’t have even access to the actual private key, just encryption and decryption functions and public key).
> 
> As a matter of fact, if one were to use HSM with routing protocol (not impossible although I guess it is mostly done nowadays with https/ipsec/ssh keys), providing private key would not be possible as only public key (and-or certificate) would even be available.

To clarify, I entirely agree with you - I think we should NOT allow extracting the private key. But my incorrect understanding of the doc was that it allowed grabbing them. I just reread it and agree that I does not allow it, only addition and deletion.

David