Markdown Version | Session Recording
Session Date/Time: 19 Apr 2024 13:00
ANIMA
Summary
The ANIMA Working Group held an interim meeting to discuss two primary topics: resolving open issues for documents currently in Working Group Last Call (WGLC), specifically focusing on CDDL definitions for the BRISKEE PRM draft, and progressing the BRISKEE Discovery draft, with a particular emphasis on the CoAP Link Format (CoLF) aspects including service name encoding, URI schemes, and the signaling of BRISKEE protocol variations.
Key Discussion Points
-
BRISKEE PRM CDDL Issues
enroll-typeDefinition: The discussion centered on defining theenroll-typefor trigger messages in BRISKEE PRM using CDDL. The goal is to allow extensible string values (e.g.,enroll-generic-cert) for different enrollment types without a formal IANA registry at this stage.- CDDL Syntax for Extensibility: Carsten Bormann advised using a CDDL "socket" mechanism for extensibility:
$enroll-type-string /= "enroll-generic-cert". This allows new documents to extend the definition without modifying the original RFC. It was also recommended to avoid unnecessary inner JSON object braces for single values. - Registry vs. RFC Updates: The group discussed whether to create an IANA registry for
enroll-typestrings. Given the expectation of only a few values (e.g.,generic-cert,app-cert), the sense of the discussion was that avoiding a registry for now and using RFC updates for new values is acceptable. An idea to include a forward-lookingenroll-app-certin the current document was cautioned against due to potential ISG pushback for undefined terms.
-
BRISKEE Discovery (CoLF)
- Objective: The Discovery draft aims to enable pledges to discover proxies, and proxies to discover registrars, and to signal various BRISKEE protocol capabilities.
- Service Name/Resource Type Encoding: Clarified the use of resource types like
brski-do-jpy(pledges discovering proxies) andbrski-rjpy/brski-rs(proxies discovering registrars) for CoLF. The consistent abbreviationJPYfor the stateless protocol was also discussed. - CoLF Multicast Example: An example was provided for multicast discovery (e.g.,
ff02::fdforbrski-do-jpy) and the expectedLink-Formatresponses. - URI Encoding Limitations: A key issue highlighted was the
Link-Formatrequirement to include the full IP address and scheme (e.g.,coaps://[ff02::fd]:8485) even when only signaling a different port on the same host. It was noted that this is an inherent limitation of theLink-Formatspecification (RFC 8288, formerly RFC 5988) and not easily circumvented without introducing non-standard extensions, which would likely face significant review hurdles. - Query Filtering: Filtering by scheme (e.g.,
coapsvs.https) in multicast CoLF queries is not directly supported by RFC 6690 for single query arguments, though it is possible when querying a CoAP Resource Directory. - CoAP over TCP/TLS: The existence of CoAP over TCP/TLS (RFC 8323,
coaps+tcp://) was confirmed, but the general sense was that it's not currently recommended for typical constrained BRISKEE deployments due to its overhead compared to UDP-based CoAP, unless dealing with very large data transfers (e.g., post-quantum cryptography). No explicit support is planned for BRISKEE at this time. - BRISKEE Protocol Variations:
- The need to signal combinations of protocol aspects (e.g., responder mode, voucher encoding, enrollment protocol) was discussed. A single string (e.g.,
RRM-CMS-EST) maintained in a registry was proposed to simplify parsing for implementations. ESTstring interpretation: It was agreed that theESTstring in variation descriptions implicitly refers to "constrained EST" when the transport context is CoAP, with the full semantics provided by the registry entry for the combined variation string.BVAttribute for CoLF: A newBV(BRISKEE Variation) attribute was proposed for CoLF to carry these variation strings.- Backward Compatibility: For RFC 8995, an empty
BVstring would signify the default. - Future Variations: Several example variations (e.g.,
RRM-Jose-CMPover HTTPS,PRM-CMP) were identified as future or currently undefined, and it was suggested they be moved to a "potential future" section in the draft rather than being presented as current specifications.
- The need to signal combinations of protocol aspects (e.g., responder mode, voucher encoding, enrollment protocol) was discussed. A single string (e.g.,
- CoAP Resource Directory (RD):
- RD can serve as a centralized discovery mechanism for proxies to find registrars.
- Registration: Devices would register their
.well-known/corecontent with the RD. The content itself (e.g., resource types,BVattributes) would be directly reusable from local discovery. - RD Discovery: The mechanism for devices (specifically proxies) to discover the RD itself is outside the scope of the current draft and typically relies on pre-provisioning or deployment-specific methods, similar to Lightweight M2M.
Decisions and Action Items
- Decision: For the BRISKEE PRM draft, the
enroll-typeCDDL definition will use the extensible socket syntax ($enroll-type-string /= "enroll-generic-cert") to allow future extensions without requiring an IANA registry for theenroll-typestrings themselves at this time. - Decision: No specific workaround for the
Link-FormatURI encoding limitation (requiring full IP/scheme/port in responses even for same host) will be pursued in the BRISKEE Discovery document due to the nature ofLink-Formatand potential IETF review complexity. - Decision: When signaling BRISKEE protocol variations via strings (e.g.,
RRM-COSE-EST), theESTcomponent will implicitly refer to "constrained EST" when the transport context is CoAP. The full semantics for such composite strings will be defined in the associated IANA registry. - Action Item (Stephan): Update the BRISKEE PRM draft with the agreed CDDL for
enroll-type. - Action Item (Toerless): Refine the BRISKEE Discovery draft to move examples of currently undefined or future protocol variations (e.g.,
RRM-Jose-CMPover HTTPS,PRM-CMP) into a 'potential future' or example section rather than presenting them as current specifications.
Next Steps
- Continue to incorporate the discussed CoLF aspects and clarifications on protocol variations into the BRISKEE Discovery draft.
- The BRISKEE PRM draft will be updated to reflect the agreed CDDL syntax for extensibility.
- Further consideration of how to best describe Resource Directory interactions for BRISKEE discovery, particularly the discovery of the RD itself, will be needed but remains a deployment-specific consideration for now.