Re: [6tisch-security] [6tisch] xxx-bootstrap

[Michael Richardson <[e-mail]>]

replying to three of your messages, and trying to repoint the thread at
6tisch-security.

peter van der Stok <[e-mail]> wrote:
    > all in favor of one approach to merge the push/pull aspects. (I have to
    > understand the protocol exchange below, but it looks feasibale)
    > I am not sure about understanding EDHOC, but may be that is not important.

I think it is.

    > I still see all the mime formats. is that phase 1?

They are HTTP content-types really.
That's part of of the phase 2.

    > Is phase 2 then switching to OSCOAP with a single format containing binary
    > encoding?

No, we would use EDHOC to create a secure session, and then we would run
our EST-CoAP-variant over that. No DTLS at all.

peter van der Stok <[e-mail]> wrote:
    > I don't think EDHOC is meant to replace EST or the anima work.

EDHOC doesn't replace EST, it replaces DTLS.

peter van der Stok <[e-mail]> wrote:
    > As discussed earlier, having a push/pull agnostic protocol would be
    > nice. (and probably feasible)

    > As stated earlier, I think that discovery and which node starts the bootstrap
    > protocol (BRSKI) is very much installation/technology dependent, and should
    > be done in another document.

I feel very much like I am the holdout for having the JCE initiate the join
protocol.

    > My intention was to reduce the coding effort of the registrar by making 1)
    > and 2) as similar as can be hoped for;
    > while payload considerations are of secondary importance.
    > My hope being that a manufacturer will deliver a registrar box supporting
    > both coap and http versions.

    > When the consensus is that adding coap is already such an extra effort that
    > code sharing between http and coap versions does not alleviate the effort,
    > then any solution for coap that concentrates on payload reduction is fine by
    > me.

I can see an implementation where a front-end requests/processes the PKCS
format messages in traditional EST and the CoAP push/pull variants, and then
feeds it into the same engine.

I think that the biggest interoperability problem that we are going to have
is in the step where the Registrar/JCE tells the Pledge what to put in it's
CSR. This is because this part probably involves doing real ASN.1 work, and
pledges will want none of that.

--
Michael Richardson <[e-mail]>, Sandelman Software Works
 -= IPv6 IoT consulting =-