[Dtls-iot] Hardware-based Random Number Generator

[Hannes Tschofenig <[e-mail]>]

Stephen wrote:

(10) 14: I want to check that the WG are really ok with the "MUST" for
h/w "quality" RNGs. I really like the idea, but it's very odd for the
IETF to have such a MUST touching on the style of implementation.  (Say
if my entire implementation of TLS is in some kind of VM? Then I
couldn't meet this ever.) I think we have to s/MUST/weasel-words/ there
tbh, sadly - it's just not our job to mandate implementation styles. The
following para though does the job nicely, so maybe s/MUST/ought
to/ is enough of a change.


In response to a similar comment from Russ I had softened the language
already. Here is the remark that Russ had:
http://www.ietf.org/mail-archive/web/dtls-iot/current/msg00273.html

I am, however, wondering whether we do ourselves a favour if we don't
state what the requirements are.

The example you have mentioned is also a bit artifical since IoT devices
don't run in a virtual machine and even virtual machines rely on
hardware features from the underlying host operating system.

I hope someone in the group can share their views.