CDNI WG Minutes IETF 118 Prague Chairs: Kevin Ma and Sanjay Mishra AD: Francesca Palombini Recording: https://www.meetecho.com/ietf118/recordings/#CDNI Chair slides: Errata 7657 https://www.rfc-editor.org/errata/eid7657: - Kevin/Sanjay: It's just a typo in the example; the normative text is correct - Francesca: Will mark the errata verified - Francesca: ACME draft waiting on a new revision - Sanjay: Talked to Frederic and he will make sure all emails have responses draft-ietf-cdni-capacity-insights-extensions - Ben Rosenblum (3:35 min) - Ben: Addressed comments from Kevin and Sanjay - Ben: Are we ready for WGLC - Sanjay: The document looks in good shape - Kevin: From the last IETF there was a question about IANA registries, what did we decide? - Ben: We decided to do registries for some and not others - Kevin: Will review the updated draft and assuming it's address we should be ready for WGLC. Will follow up on list. draft-ietf-cdni-https-delegation-subcerts - Christophe Neumann(3:40 min) - Christophe: Addressed comments from SecDir early review: encrypting the optional private key - Christophe: Are we ready for WGLC - Sanjay: Make sure to confirm on the mailing list with the SecDir reviewer. The draft looks good. Sent comments to the list. - Kevin: I agree, have Mike confirm the list. Otherwise the draft looks good once Sanjay's comments are addressed. After that we're probably ready for WGLC draft-ietf-cdni-ci-triggers-rfc8007bis - Sanjay Mishra (3:45 min) - Sanjay: Addressed IANA comments, working on comments from Alan, need to address comments from Kevin, should have a new draft in December ready for WGLC - Kevin: I still need to review the examples and the IANA sections. If we can address Alan's comments as a separate extensions draft that would be better, unless fundamental changes to the protocol need to get into base draft - Sanjay: Setting up working sessions with Alan to resolve draft-ietf-cdni-protected-secrets-metadata - Ben Rosenblum (3:50 min) - Ben: Addressed Kevin's comments, still needs some sequence diagrams - Ben: Open question: Should we keep FCI objects? There's no precedence for reusing MI objects for FCI. - Kevin: Yes, we didn't need it with the original set of obejcts, but I prefer less redundancy - Ben: Agree with less redundancy, just wasn't clear if that was acceptable - Glenn: FCI objects make it clearer - Ben: Will take the question to the list draft-ietf-cdni-edge-control-metadata - Will Power/Glenn Goldstein? (3:55 min) - Glenn: Addressed Kevin's comments - Glenn: Ready for WGLC? - Kevin: I need to review updated draft and encourage everyone to do the same. Once reviewed, we can go from there. draft-ietf-cdni-edge-control-metadata - Alfonso Siloniz (4:05 min) - Alfonso: Addressed Kevin's comments - Alfonso: Sent responses to open question about mts/sfd - Kevin: I didn't get a chance to review the updated draft. Will respond to the questions on the list draft-power-metadata-expression-language - Will Power (4:15 min) Glenn presenting for Will having technical difficulty - Glenn: Work on the MEL has gone through many rounds of review in SVTA, now looking for adoption in CDNI - Glenn: Draft defines the expression language, one MI object, and one FCI object - Kevin: FCI object is definitely within scope, but the MEL does not seem CDNI-specific, should it be in HTTP - Chris Lemmons: It's not entirely generic and not CDNI-specific. It is intermediary-specific, and as such, CDNI might be more excited to do the work - no comment on whether it fits in the charter. - Kevin: Sanjay and I need to talk with Francesca about how it fits in the charter or if we need to modify it, or if we should shop it around to other working groups - Francesca: No additional comment at this time - Sanjay: Agree, need to look and see if the motivation for MEL is a fit for CDNI - Yoav Gressel: The language is generic, but it is important to CDNI use case. It is required for CDNI to work. Could add metadata properties, but this language would open up the domain. - Chris: This language is required for different intermediaries to communicate in an interoperable way; it is a protocol agreement, so it belongs in CDNI - Alan Arolovitch: It is more broadly applicable and probably could be extended and used beyond just CDNI, but there is interest in building/adopting this and the proxy cache scope is critical here - Rajeev RK: Votes to include this work - Alfonso: If not this language, we still need this functionality to configure the dCDN from uCDN; making it usable for other applications is good, but it's useful for CDNs - Kevin: There's cleary enthusiasm for the work. Sanjay and I just need to discuss how it fits in the charter - Glenn: The language is generic, but proposed variables and functions are CDNI specific draft-goldstein-processing-stages-metadata - Glenn Goldstein (4:30 min) - Rajeev: This is powerful for lightweight implementation of other standards like TraceData/ProxyStatus/CMCD/CMSD. - Sanjay: Not clear on how processing stages are used to implement CMCD - Rajeev: Example, adding a new variable to CMCD required modifying the player; could do it with MEL/processing stages - Glenn: Could synthesize CMSD headers - Rajeev: Could send unique headers to clients (e.g., for calculated bandwidth) - Glenn: Looking for WG adoption - Kevin: Have a similar question to MEL wrt how it fits in the charter - Kevin: Show of hands on support for processing stages: 14 yes - Kevin: Sanjay and I will discuss how it fits in the charter and continue the disucssion on the list draft-rosenblum-cdni-logging-extensions - Ben Rosenblum (4:45 min) - Ben: RFC7937 only defines one specific format and is insufficient - Ben: SVTA did a survey of operators to get requirements, e.g., modern formats, archives, transports and updates for modern privacy regulations - Rajeev: Draft is backwards compatible with RFC7937 - Rajeev: I support the draft. With broader adoption, automated transfer becomes more necessary. Useful for billing - Ben: The work will happen either way. If not by CDNI, through proprietary protocols - Rajeev/Ben: - Kevin: poll on who as read the draft: 7 yes, 7 no - Ben: SVTA work will continue; if CDNI WG is interested, we can see how to collaborate - Kevin: I need to read the draft and encourage others to read the draft - Kevin: The issues raised are important and the original draft lacks capabilities needed today - Sanjay: Please read the draft and send comments to the list. It will help gauge interest Cache Management - Alan Arolovitch (5:00 min) - Alan: Need more automation of pre-position/invalidation/purge using triggers - Chris: Are you familiar with cache groups and cache invalidation API work in HTTPbis? There is a large overlap. Is this work for CDNI or for a CDN in general? https://datatracker.ietf.org/doc/draft-ietf-httpbis-cache-groups/ https://datatracker.ietf.org/doc/draft-nottingham-http-invalidation/ - Alan: Is it for individual cache management? - Chris: Includes discovering API endpoints and ask an entire CDN to purge a URI - Alan: This extends existing CI/T, so need to harmonize that with HTTPbis work - Kevin: Is something not supported by URI pattern matching or is URI pattern matching just too cumbersome and grouping is more efficient? - Alan: Both. Not feasible Example: Manifest + chunks + subtitles + thumbnails are not easily pattern matched. Makes automation easier if upstream can publish tags with content. Named footprint - Alan Arolovitch (5:15 min) - Alan: Named footprints come before ALTO - Sanjay: Draft has expired. Suggest resubmitting v1 and we can discuss on the list