[Opendnssec-develop] True Random Number Generator

John Dickinson jad at jadickinson.co.uk
Thu Jan 8 12:36:43 UTC 2009


On 7 Jan 2009, at 14:33, Rickard Bondesson wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA256
>
>> RSA is a different matter.  With that, random material is
>> only needed when generating keys.  Unless you are signing
>> loads and loads of domains you need nothing to speed up
>> random generation for that, I'd imagine.
>> A _good_ source is still advisable of course, and hardware is
>> so incoherent it produces far better generators than software.
>>
>> When signing for DNSSEC, the choice between RSA and DSA is easy:
>> - RSA keysizes can be increased as security demands;
>> - RSA needs no masses of random material when in signing operation;
>> - RSA validates much quicker than DSA.
>
> I agree that RSA is a good choice. Currently there is no support for  
> DSA in SoftHSM. Hardware TRNG is thereby, as for OpenDNSSEC, not a  
> high priority, just a nice thing but would need device dependent  
> code since there is no standardized interface. I will make it easy  
> to extend the SoftHSM with such code.

I did use one of those Araneus things once. I seem to remember it  
being easy to create a file full of random data. Would it be better to  
have the Araneus appear as an alternative /dev/random device that you  
point the softHSM at? Or am I completely misunderstanding?

One other thing that I thought would be good is if the softHSM can be  
complete enough to work with an OpenSSL pkcs11 engine (like the OpenSC  
one). I know we don't want that for OpenDNSSEC but it might be a good  
feature to have. WDYT?

Thanks
John



More information about the Opendnssec-develop mailing list