Hi Mike, A few remarks inline.
On 5/14/2019 7:29 PM, Hannes Tschofenig wrote: > Hi Paul, > > My understanding from reading the draft text was that the "cost" was actually > talking about "energy cost" rather than "monetary cost". > The monetary cost may also be interesting. > > It is difficult to judge the extra cost of a RNG in an MCU because > (a) you rarely find an MCU with and without MCU (keeping all other > features the same), "with and without RNG"? > (b) even if you find one there are other factors that impact the cost > (such as popularity of a particular MCU), > (c) RNG features are often provided with other features (such as > SHA256 and AES in hardware), and Um... you usually need one or the other of these to do a good implementation of a DRBG and hardware versions tend to be less "costly" energy wise than software ones. > (d) cost and price of an MCU are different aspects. That have no bearing on the protocol design past a certain set of minimums. I'm beginning to think its time for "minimum requirements for internet hosts - 2020 edition" RFCs. We seem to be back in the same set of arguments that ANYTHING should be able to be an internet device albiet with even lower thresholds than we were a few years ago. ACE is pushing those boundaries, but we seem to be chasing "cost -> 0" [Hannes] Good that you are supporting my argument. I believe this is the first time you agree with me. All good points, but ignoring the fact that needs drive the protocols which drive the MCU development. NOT MCUs limit the protocols which limit the needs. [Hannes] You should join the EDHOC discussion where you can hear a lot about how MCUs and networks limit (security) protocols. In other words, if the community (through protocol definitions) doesn't require the RNG, then the MCU developers won't include it. [Hannes] Luckily silicon vendors already include RNGs in MCUs. That mission got accomplished. With respect to RNG energy costs - the argument that they are too expensive energy wise is pretty bogus. [Hannes] Completely agree with you. Luckily that text part got changed in the latest version of the draft as a consequence of the recent mailing list discussion. RNGs generally have two parts - a TRNG or entropy source, and a DRBG seeded from the entropy source. The energy costs for the DRBG are microscopic. The energy costs for a TRNG are related more to the time needed to initially fill the entropy pool (e.g. getting the noisy diodes making noise, getting the ring oscillators ringing etc) than to keeping it running to get a few thousand bits. Doing this say once a year to seed the DRBG isn't going to kill the lifetime of a battery device. [Hannes] Well said. I had hoped that someone would go into the details and you did. Btw, did you watch my webinar about RNGs https://www.youtube.com/watch?v=xQF2U6AkSjI ? Ciao Hannes IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you. _______________________________________________ Ace mailing list Ace@ietf.org https://www.ietf.org/mailman/listinfo/ace