Sender: minutes-request@ietf.CNRI.Reston.Va.US From: Leslie Daigle Date: Thu, 11 Jul 1996 17:29:13 -0400 To: minutes@CNRI.Reston.Va.US Subject: Minutes for Montreal URN BOF Attached are the text minutes for the URN BOF in Montreal. The electronic versions of the presentation slides can be found at: Session slides: ftp://ftp.bunyip.com/research/ietf/urn-bof/chairslides.ps Framework presentation slides: ftp://ftp.bunyip.com/research/ietf/urn-bof/framewk.ppt NAPTR presentation slides: ftp://ftp.bunyip.com/research/ietf/urn-bof/naptr.ppt Sollins/Girod presentation slides: ftp://ftp.bunyip.com/research/ietf/urn-bof/solgir.ppt Let me know if there are any problems. Thanks! Leslie. ================================ Uniform Resource Names (URN) BOF CURRENT MEETING REPORT Reported by Dirk van Gulik, Joint Research Centre of the European Communities, and Leslie Daigle, Bunyip Information Systems Inc. ---------------------------------------- Purpose of the BOF ---------------------------------------- This BOF was to report on the URN work that has been carried out since the Dallas IETF meeting in December, 1995. Three Internet-Drafts had been prepared that describe some of this work (see below). Given that this work represents significant progress towards addressing many of the URN requirements, a possible plan was outlined for finalizing this work and determine if there is consensus for continuing it within the context of the IETF. Related Internet-Drafts: draft-daigle-urnframework-00.txt draft-daniel-naptr-01.txt draft-girod-urn-res-require_00.txt ---------------------------------------- Agenda ---------------------------------------- Presentations: . Welcome, context . Presentation of framework document . Presentation of the NAPTR document (draft-ietf-daniel-naptr-01.txt) Forward motion: . Next steps -- proposed charter for a WG . Presentation of the Girod/Sollins paper -- suggestions for improvements to the framework/NAPTR stuff . Discussion of other proposed improvements for the current proposals. . Closure on WG charter plan ---------------------------------------- Notes ---------------------------------------- The attached slides carry the details of individual presentations. 1. Welcome: Leslie Daigle From consensus in the room, an new mailing list will be set up for this phase of URN discussions: urn-ietf[-request]@bunyip.com This segment included a short introduction and expose on the scope of the current URN framework (the proposal), and the NATPR draft (a proposed implementation of one of the components). In particular: what URNs are not going to solve. It was noted that the Framework document acknowledgements section was incomplete, as Bill Arms's name was not listed -- an unfortunate editorial mistake. 2. Framework Document: Patrik Faltstrom This presentation outlined the proposed framework -- as it existed from the Dallas BOF and a few refinements since. There were some questions about exact NA extraction were deferred to the NAPTR discussion. There were also questions regarding choices for this specific solution: in particular whether the top-level registry should see the full URN (which would make some optimization and efficiency possible). It was pointed out that in the light of various privacy/denial-of-service attacks as well as efficiency or caching that this was considered but not selected. 3. NAPTR: Ron Daniel See draft -- just one particular way of implementing the registries of the framework. Some points were unclear in terms of who changes which NAPTR/SRV record. Regarding special port/protocol combinations, this is, or should be in the SRV draft. If not, in the interest of caching? This should be added to the NAPTR in some way? There were questions as to whether 'executing' a regex is not a security problem. The three required DNS lookups were questioned, given historic proposals. The current proposal, however, does offer more functionality, records can be cashed more effectively and the SRV record allows for more optimizations when combined with shortcuts in DNS. It is noted that security is an issue and thread models are to be identified and which can be addressed. 4. Next Steps: Leslie Daigle Given the proposal on the table (as outlined in the earlier presentations), a draft Working Group charter was floated, to see if there was consensus that a group could be formed to do useful work (see slides for details). With regard to the specific charter and proposed milestones, there were Concerns that the one resolution scheme and one name space as proposed is too little; but a single year working group is limited in scope. Proof that grandfathering in a single name space is needed (e.g., FPI's or IS?Ns) and of course, examples of other N2L, etc resolutions, such as cri or enterprise numbering (in another paper). Furthermore, a separate URN syntax document is needed. It was stated that the framework document will refer to the existing URN requirements RFC, but it will not replace it. The operational impact on DNS can be serious and should be considered within this working group. However a complete analysis belongs to the SRV and DNS communities as well. However during design these issues have been taken into account and it is expected that the type of load introduced should be similar and match well with current optimizations. 5. Improvements: Karen Sollins, Lewis Girod This presentation kicked of the constructive criticism of the NAPTR proposal. The related Internet-Draft (draft-girod-urn-res-require_00.txt) was written before the other two Internet-Drafts were circulating publicly, so this presentation focused on issues that are still open after integrating all material. A compulsory gatewaying, such as a HTTP client defaulting to say HTTP-proxy, could be useful for requiring minimal changes when the URN infrastructure changes, but it is difficult to force vendors to do so. Also firewalls and DNS are an issue. Some design issues around DNS were raised; the NAPTR use might be at odds with the bulk security signing model proposed in DNSSEC. An important clarification, which might have to be more emphasized in the draft, is that the rewrite rule applies to the original URN, the output of one rule is not used as the input for another rule. It was questioned whether this kind of a regular expression rewrite rule is enough to accommodate all needs; but it seems to give enough control over granularity to implement most management systems foreseen by the designers. Some of the more complicated steps might have to be done by referrals of the final resolving server. However in the light of the requirements they are felt as needed. Whether the original requirements were good is another question -- should that document be re-written? 6. Discussion It seems that most problems are deferred to the resolution protocols. It is therefore important to get the Framework document finished quickly. There is a lot of infrastructure to build. The framework document should borrow some of the flesh from the Girod/Sollins draft; it gives more compelling reasons of why URNs are a Good Thing. 7. Closure WG Charter plan There was consensus in the room that this proposal represents a reasonable path to tackle the URN problem. The charter will be firmed up and proposed to the IESG. ------------------------------------------------------------------------------ Leslie Daigle "Learn and live." Vice President, Research Bunyip Information Systems -- ThinkingCat (514) 875-8611 leslie@bunyip.com ------------------------------------------------------------------------------