Re: [Ace] draft-selander-ace-eals vs. draft-vanderstok-ace-coap-est

[Göran Selander <[e-mail]>]

Hi Hannes, 

It is interesting to note that during the last 6 months there seems to be
very little activity to encourage discussion or reviews of the drafts
which actually are in scope of the current ACE charter. For example, what
do people think about the overlaps between the two publish-subscribe
profiles (draft-sengul-kirby-ace-mqtt-tls-profile vs.
draft-palombini-ace-coap-pubsub-profile)? Instead, you as co-chair of ACE
encourage the WG to discuss two drafts about certificate enrolment, a
topic which is not even mentioned in the current charter.

As co-author of one of the drafts in the subject of this mail I of course
welcome the discussion, but it is important that this thread does not to
distract the ACE WG from working according to its charter on its missed
milestones.

Now for the topic of comparing the certificate enrolment drafts.
draft-vanderstok-ace-coap-est proposes a very natural and significant
optimization of EST adapted to the established security setup for CoAP,
and I fully support that. There are overlapping authors between the
drafts, so clearly the drafts should not be seen in opposition to each
other.

My view of the potential contribution with draft-selander-ace-eals (which
is a first sketch we made in the spring, and which I recently stepped in
revision just to keep alive) is twofold:

1. It discusses the generalisation of EST to application layer security.
The enrolment procedure in EST is in principle not dependent on what layer
authentication takes place, provided there is security end-to-end between
the endpoint making the enrolment request and the endpoint providing the
certificate. As we know, there are common IoT settings where security on
transport layer does not go end-to-end because of gateways or proxies or
because of change of underlying layers, which is the reason for proposing
this complement on the application layer. As to the actual enrolment
procedure, it may well be the same in both cases of transport layer
security or application layer security.

2. In a second independent component (section 3.2), the ACE framework is
applied to authorise and provide keys to the endpoints involved in the
enrolment, after which the very same enrolment procedure can take place.
This shows a more lightweight key establishment than with a key exchange
protocol (such as the DTLS handshake or EDHOC) with fewer and smaller
messages, and less public key operations, all of which are favourable
properties in constrained environments.

The implicit question posed by draft-selander-ace-eals is the following:
If we are considering one IoT variant of EST
(draft-vanderstok-ace-coap-est) should we also consider other variants
using the same enrolment procedure, which can be applied to a wider range
of IoT use cases and/or which are more favourable in settings with
constrained IoT devices?



Göran




On 2017-09-13 17:42, "Ace on behalf of Hannes Tschofenig"
<[e-mail] on behalf of [e-mail]> wrote:

>Hi all,
>
>in previous IETF meetings we had presentations on these two documents
>and it appears that there is an overlap. So far we haven't had a lot of
>discussions on these proposals on the list but since there seems to be
>interest from the folks attending the IETF meetings I am recommending to
>have a discussion about the direction we should go with this work.
>
>Any thoughts?
>
>Ciao
>Hannes
>
>_______________________________________________
>Ace mailing list
>[e-mail]
>https://www.ietf.org/mailman/listinfo/ace