Re: [Mud] [Iot-onboarding] updates to diagram

[Michael Richardson <[e-mail]>]

Kent Watsen <[e-mail]> wrote:
    > The diagram has the label "NETCONF", but this is confusing...while
    > produced by the NETCONF WG, the solution itself does not depend on the
    > NETCONF 
    > protocol (though it MAY bootstrap a secure NETCONF session) From a protocol
    > perspective, it might be more appropriate to have HTTP, or REST, or even
    > RESTCONF. That said, I think the label "SZTP" is best, as that is the
    > acronyms used in the RFC.

For lack of a better term, I'll change "NETCONF" to "SZTP (netconf)"

    > I'm unsure about what is implied by the label MASA. If just the acronym as
    > defined in the voucher draft, then there should be an arrow pointing to
    > the SZTP box as well. However, if it implies a functional component
    > (protocol API), then there should be another box called "OVIS"
    > (ownership voucher 
    > issuance service) that points to the SZTP box.

My intention of the MASA box is that it involves itself in the middle
three items, but not the left-most "ultra-contrained bootstrap"

I have added an equivalent box and marked it "OVIS"

    > Lastly, the bottom row appears to capture statefulness of the
    > connection to MASA.

No, it's not connection to the MASA, the MASA box is just at the bottom
in order to not overlap.   The statefulness is intended to be the
transport that is under the boxes above.
That is, BRSKI is:

     JSON-voucher -over-
     CMS          -over-
     TCP          -over-
     circuit-level

while 6tisch-zero-touch is:
      CBOR              -over-
      COSE              -over-
      EDHOC/LAKE        -over-
      OSCORE            -over-
      stateless CoAP proxy

    > FWIW, the SZTP solution doesn't necessitate a MASA
    > at the time of the 
    > bootstrapping event. To be clearer, the pledge MAY send a nonce to a local
    > SZTP server, which MAY in turn use that nonce to retrieve an ephemeral
    > voucher from a MASA/OVIS system.

Agreed... I'm in general assuming that it could be USB/sneaker-net.
I don't know how to represent this well.


-- 
]               Never tell me the odds!                 | ipv6 mesh networks [ 
]   Michael Richardson, Sandelman Software Works        | network architect  [ 
]     [e-mail]  http://www.sandelman.ca/        |   ruby on rails    [