From: Taps <[email protected]> on behalf of Éric Vyncke via Datatracker <[email protected]> Sent: 28 August 2023 10:47
Éric Vyncke has entered the following ballot position for draft-ietf-taps-arch-18: Discuss When responding, please keep the subject line intact and reply to all email addresses included in the To and CC lines. (Feel free to cut this introductory paragraph, however.) ---------------------------------------------------------------------- DISCUSS: ---------------------------------------------------------------------- # Éric Vyncke, INT AD, comments for draft-ietf-taps-arch-18 Thank you for the work put into this *NEAT* document (private joke). It is easy to read and is an important piece of work required to deploy new transports. Please find below one blocking DISCUSS points (mainly to have a discussion, do not worry too much), some non-blocking COMMENT points (but replies would be appreciated even if only for my own education), and some nits. Special thanks to Michael Welzl for the shepherd's detailed write-up including the WG consensus and the justification of the intended status *even* if I disagree with the intended status (see below my DISCUSS point). Other thanks to Bernie Volz, the Internet directorate reviewer (at my request), please consider this int-dir review: https://datatracker.ietf.org/doc/review-ietf-taps-arch-18-intdir-telechat-volz-2023-08-25/ (minor nits) I hope that this review helps to improve the document, Regards, -éric # DISCUSS As noted in https://www.ietf.org/blog/handling-iesg-ballot-positions/, a DISCUSS ballot is a request to have a *discussion* on the following topics: ## Intended status This is only to have a public discussion (over email before the telechat or during the IESG telechat), I intend to ballot either NoObj or Yes after this discussion. The shepherd's write-up writes that the intended status is "proposed standard" per TAPS WG charter and I do not see anything related to an architecture document in the charter and even less about its intended status. Moreover, most IETF architecture documents are 'informational'. <tp> Which I see repeatedly as generating more work for the IESG:-( Documents such as YANG models and protocol specifications need a Normative Reference and the information thye need is specified in the architecture model which then has to be flagged and discussed by the IESG as a downref. Better to make architectures (and informational models) Proposed Standard in the first place. Tom Petch See also my comments about section 3.1 ---------------------------------------------------------------------- COMMENT: ---------------------------------------------------------------------- # COMMENTS ## Anycast address This document differentiates between unicast and multicast addresses, but should there be a specific case of anycast addresses ? ## Section 1.4 I am not a transport expert but I would have included the transport protocol in `Socket: The combination of a destination IP address and a destination port number [RFC8303].` ## Section 2 Should 'DNS' be included in `system-provided stub resolver` ? Figure 1 & 2 are nice but, please, add a references to them in the text. In `it describes how implementations can use multiple IP addresses` isn't it hidden usually to the application ? ## Section 2.3 In `The Socket API for protocols like TCP is generally limited to connecting to a single address over a single interface.` should there be a mention of one or several 'source' IP addresses ? Should 'address' be qualified by 'IP' (as opposed to a DNS name or "Internet address" aka URL)? ## Section 2.4 How can a (nice) informational RFC 8170 "requires" in `incremental deployability [RFC8170] requires coexistence`. Suggest to use "recommend" or something similar to avoid confusion. ## Section 3.1 The presence of normative BCP14 terms ("SHOULD", ...) in an architecture document looks weird to me (see my DISCUSS point above). Is this document an 'architecture' document or an 'architecture and requirements' one ? ## Section 3.3 What is the exact meaning of 'safely' in `Equivalent Protocol Stacks can be safely swapped or raced in parallel` ? ## Section 4.1 s/Establishment (Section 4.1.4) focuses on the *actions* that an application *takes on* the connection objects/Establishment (Section 4.1.4) focuses on the *requests* that an application *sets to* the connection objects/ as it is not really the application doing those actions ? ## Section 4.1.1 Please state the obvious: a multicast endpoint can only be a destination endpoint. ## Section 4.1.3 Do the security parameters include DNS resolution security parameters ? E.g., mandatory use of DNSSEC or DoH? ## Section 4.1.5 Unsure whether the sentence `Messages are sent in the payload of IP packet` is really useful. Suggest to remove it. ## Section 4.2.2 Suggest to mention RFC 7556 in the discussion about different local addresses (interfaces?) and DNS resolvers. # NITS ## Section 2 Is a capitalised "Connections" required in `the interface for an application to create Connections and transfer data` ? Or should there be a text in the glossary section about the use of capitalised terms ? ## Section 2.1 s/all interaction using the Transport Services API is expected to be asynchronous/all interactionS using the Transport Services API ARE expected to be asynchronous/ ? _______________________________________________ Taps mailing list [email protected] https://www.ietf.org/mailman/listinfo/taps _______________________________________________ Taps mailing list [email protected] https://www.ietf.org/mailman/listinfo/taps
