CDNI Working Group Minutes IETF-88, Vancouver, Canada - Chaired by Francois Le Faucheur and Daryl Malas - Meeting raw notes captured by Matt Caufield and Jan Seedorf, edited by Francois Le Faucheur & Daryl Malas - Slides accessible at: https://datatracker.ietf.org/meeting/88/materials.html#wg-cdni Thursday, November 7, 2013, 15:20-17:20, Georgia A =================================================== - about 50 people in the room, a few people on WebEx, no jabber scribe Introduction and Agenda (WG chairs) --------------------------------------------------------------- - Introduction by the WG chairs, and Note Well statement. - Agenda review - Charter Update: * draft new charter distributed on the list, under IESG review * "URI Signing" already added to list of deliverables * Status/Dates already updated - Published RFCs: * RFC 6707: CDNI Problem Statement (Informational) * RFC 6707: CDNI Use Cases (Informational) * RFC 6983 : Models for HTTP-Adaptive-Streaming- Aware CDNI (Informational/Independent Submission) - Document Update and progress against the charter milestones * cdni-requirements: in "IESG Last Call" state * cdni-framework: Daryl will prepare Write-Up and request publication * other documents, to be discussed in rest of the WG meeting - Documents beyond the charter: * Redirection Loop Prevention in Iterative redirection: see discussion below * Control Interface/Bootstrapping: see discussion below - Instructions to all document editors: * Run I-D Nit tool * Align to final interface names * Refer or use the expanded reference model in cdni-framework CDNI Metadata, draft-ietf-cdni-metadata-03: Matt Caulfield ---------------------------------------------------------- - Jan Seedorf, Matt Caulfield, Kevin Ma, Jon Peterson : discussion on IANA considerations and why footprint sub-registry is in metadata document, agreement that footprint sub-registry is fine to have in metadata document; - Francois: question to authors: Does the document now cover what is necessary for base HTTP delivery? - Matt: only extra bit needed for base HTTP delivery is more flexibility in Query String handling - Francois: document needs to be revved up to: * fix IANA text * add flexibility in Query String handling - Daryl asks for one more thorough review (on next rev) before going to WG Last Call. - Vijay Gurbani: agree to do a review (within 2 weeks of next rev issued) - Daryl: once this review is completed and comments addressed, we will go to WG Last Call. FCI Semantics, draft-ietf-cdni-footprint-capabilities-semantics-01: Jan Seedorf ------------------------------------------------------------------------------- - Jan Seedorf: All open issues have been closed. - Jan Seedorf: Added IANA consideration section for capabilities. - Francois Le Faucheur: Can an informational RFC define IANA registry? - Kevin Ma: Yes, decided in Berlin (based on Jon Peterson's guidance) that this is permissible. - Spencer Dawkins: Creating a registry in an information RFC which requires standards action may not be permissible. - Jon Peterson: Consider using "IETF Review" as the process for adding new IANA values. - Spencer Dawkins: RFC5226 - well known IANA policies, not required, but should be reviewed. FCI Advertisement with ALTO, draft-seedorf-cdni-request-routing-alto-05: Jan Seedorf ----------------------------------------------------------------------------------- - Jan Seedorf: Currently two solution proposals: ALTO and Web Services based. - Jan Seedorf: ALTO solution supports a level of indirection between footprints and capabilties, which provides an optimal encoding. - Jan Seedorf: ALTO still working on defining incremental updates, could use JSON PATCH in the meantime. - Kevin Ma: Difficult to encode updates when spliting PIDs into multiple groupings. - Scott Leibrand: Does this proposal enable geo-fencing? - Jon Peterson: Expecting this encoding to be used for incremental updates - not "South of France" contains these IP addresses. FCI Advertisement with Web Services, draft-ma-cdni-capabilities-04: Kevin Ma ---------------------------------------------------------------------------- - Kevin Ma: Updated interface names, aligned to semantics draft, switched to JSON - Kevin Ma: Added ability for incremental updates. - Kevin Ma: Should we select a single FCI protocol or multiple? - Daryl Malas: Every time this topic comes up, usually a long discussion and no conclusion. - Jan Seedorf: ALTO is more mature, for example error code. Both proposals have similar encoding for footprint and capabilties. - Jon Peterson: Only really starting now to compare the proposals. Not advocating for either. Do not mind choosing both. The protocols are really just wrappers if they are tunnelling the same objects. - Vijay Gurbani: Struggled in ALTO to distinguish between HTTP error code and payload error code. - Jan Seedorf: Would encourage working group to please read documents. - Daryl Malas: Helpful for someone to read both and write up a comparative analysis. - Matt Caulfield: volunteered to write-up analysis. - Daryl: OK, Matt please prepare this analysis. Jan and Kevin, please provide input to Matt. CDNI Logging, draft-ietf-cdni-logging-08: Iuniana Oprescu --------------------------------------------------------- - Iuniana Oprescu: 2 informal meetings and 3 revs published since previous IETF meeting - Iuniana Oprescu: all open issues addressed - Iuniana Oprescu: Removed non-repudiation and will extract to a separate document. - Iuniana Oprescu: Added statement on DOS attacks. - Iuniana Oprescu: Updated mandatory/optional usage. - Iuniana Oprescu: Cleared up "Mandatory to Implement" vs "Mandatory to Use" - Iuniana Oprescu: Proposal to move to WG last call? - Jan Seedorf: CDN Logging capabilities are all optional? - Kevin Ma: Capabilities doc still includes editors notes on this topic. - Daryl Malas: Kevin, can you review and send out comments to the list? - Kevin Ma: Yes. - Daryl Malas: Can we proceed with the logging document if other interfaces are not yet complete? - Kevin Ma: Yes. If registries are well-defined, then other interfaces can add to logging registries. - Jon Peterson: +1 - Francois Le Faucheur: Logging document only defines fields for HTTP delivery and transport for logs. Should be unrelated to other interfaces. - Ben Niven-Jenkins: +1 - Daryl: alright, so let's have Kevin conduct a review within 2 weeks, and then once these review comments are addressed we can move to WG Last Call. CDNI URI Signing, draft-leung-cdni-uri-signing-03: Kent Leung ------------------------------------------------------------- - Kent Leung: Potential Issue with standardizing structure of a URI (see draft-ietf-appsawg-uri-get-off-my-lawn). Considering several resolutions and have a proposal. Meeting with Mark Nottingham has been scheduled following this meeting. - Kent Leung: Should level of security be maintained between CDNs? i.e. can hash also switch from SHA1 to MD5? - Kent Leung: Adopt this draft as a WG document? - Daryl Malas: Hold off until after discussion with Mark and general approach wrt draft-ietf-appsawg-uri-get-off-my-lawn is established. Routing Request Redirection for CDN Interconnection, draft-ietf-cdni-redirection-01: Ben Niven-Jenkins ------------------------------------------------------------------------------------------------------ - Ben Niven-Jenkins: Mainly editorial changes since last revision. - Ben Niven-Jenkins: Hesitant to define an entire list of error codes, piece of draft requires more work. - Francois Le Faucheur: Could define a small set of basic error codes and create a registry for extensibility. - Daryl Malas: Concerned with not enough detail in this area. - Scott Leibrand: Sufficient to use existing HTTP error codes unless we expect uCDN to do something special with a more elaborate error scheme. - Ben Niven-Jenkins: Draft also needs to be aligned with URI signing draft. - Kent Leung: yes, URI signing has some implications on RRI. - Kevin Ma: HTTP 400 normally used for invalid input (malformed request) - Ben Niven-Jenkins: Three types of error: 1) Propagate an error to user agent, 2) Malformed request, 3) Internal error (e.g. unknown hostname) - Francois Le Faucheur: What would it take to accelerate the progress on this doc so it can meets its charter deadline? would additional editing cycles help? - Ben Niven-Jenkins: there is no unsurmountable known issues, just needs work and editing cycles. - Daryl Malas: Looking for volunteers for co-editing. Request interface for CDN Interconnection, draft-choi-cdni-req-intf-01: Francois (on behalf of Taesang Cho) ----------------------------------------------------------------------------------------------------------- - Francois Le Faucheur: New version reflects decision from IETF87 to: * include loop prevention staff related to Recursive mode in RRI * have loop prevention staff related to Iterative mode in this separate document (covering all aspects including request interface). CDNI Control - Triggers, draft-ietf-cdni-control-triggers-01: Rob Murray ------------------------------------------------------------------------ - Rob Murray: Allows uCDN to trigger events in the dCDN like invalidation or prepositioning - Rob Murray: Adopting proposal to conform to other standard for links in JSON. Should consider modifying other drafts, like Metadata. - Matt Caulfield: How concrete is this new syntax as a proposed standard? - Ben Niven-Jenkins: It's a de facto standard, and is cleaner way of representing links. - Francois Le Faucheur: Would it introduce a dependency on Internet-Drafts outside of our control? - Ben Niven-Jenkins: We can avoid that by specifying something consistent with the de facto standards, without relying on a reference to other documents. - Matt Caulfield: Given that I think it is okay to adopt for Metadata. CDNI Control - Initialization and Bootstrapping, draft-choi-cdni-control-init-bootstrapping-02: Taesang Cho ----------------------------------------------------------------------------------------------------------- - Taesung could not attend IETF-88 - Francois Le Faucheur: the draft proposes a complement to the Control Interface, which covers bootstrapping/initialisation of CDNI interfaces. See I-D. CDNI Rate Pacing, (draft-caulfield-cdni-rate-pacing-00): Matt Caulfield ----------------------------------------------------------------------- - document defines a CDNI solution to allow the uCDN to control the rate at which the dCDN delivers content on behalf of the uCDN. This is so that in turn, the uCDN can control overall delivery speed on behalf of the CSP. - document discusses impact on all interfaces - Scott Leibrand: Why would you need an additional "ac-rate" logging field, when it seems you can compute something quite similar from existing logging fields? - Matt: because it does not give the exact same result since it would factor in the acquisition start-up time - Scott Leibrand: but acquisition time is usually negligible since delivery takes place in pipeline mode with acquisition. - Francois Le Faucheur: I also don't see that it is justified to define a new field. Please discuss on the list the rationale. - Francois Le Faucheur: As mentioned on the list, I don't quite buy the security risk that is brought up in security considerations. Meeting closes --------------