Minutes of the IETF DRINKS WG meeting during IETF 81 ==================================================== Date/Location: TUESDAY, July 26, 2011, 1520-1700pm Quebec City, Quebec, Canada Agenda: http://www.ietf.org/proceedings/81/agenda/drinks.txt Slides: https://datatracker.ietf.org/meeting/81/materials.html http://www.ietf.org/proceedings/81/drinks.html Audio recording: http://www.ietf.org/audio/ietf81/ietf81-2103-20110726-1256-pm.mp3 Meetecho recordings: http://www.meetecho.com/ietf81/recordings Mailing List archive: http://www.ietf.org/mail-archive/web/drinks/current/maillist.html WG Chairs: Sumanth Channabasappa, Alex Mayrhofer RAI ADs: Gonzalo Camarillo, Robert Sparks 1) WG Session Welcome & Administrivia (Chairs, 5m) -------------------------------------------------- Link: http://tools.ietf.org/agenda/81/slides/drinks-4.ppt - Scribes were identified: Syed Ali (note taker) and Ken Cartwright (backup while Syed presents), Kevin Fleming (Jabber) - Alex presented the Note Well and administrivia - Agenda is presented. - Gonzalo--Notes smaller attendance and lack of activity on the list. Advises on wrapping the work. Future work shall be handled in DISPATCH. XCON and SIMPLE are examples of where WG is shutting down, and new work is being considered within DISPATCH. In short, new work should probably be picked within DISPATCH. 2) WG Status Chairs (10m) ------------------------- Link: http://tools.ietf.org/agenda/81/slides/drinks-4.ppt - Document status of WG items - Alex provided an update regarding the three WG documents; in summary: > the use cases document will be presented for publication soon > the protocol and transport document should be completed by Sep. - Protocol document: Dean submitted comments that mainly arrived from implementers. - transport document: comments from Manjul - Transport: Peter St. Andre posted WGLC comments as well. And therefore, there will definitely be updates to the transport document - David asks about bringing in additional work such as building another transport based on RESTful principles. - Gonzalo clarifies that DISPATCH would be the way - David also asks brings up OpenSPPP work and addressing subsequent issues that may arise. And how closing the WG may be an issue. - Gonzalo said that the list can remain open and can be used for further dialog in order to address any potential issues that may arise later. - (Gonzalo) provided many options: go dormant, shutdown the WG and take new work to DISPATCH, AD-sponsored changes etc. - Alex: milestones and status. 3) Protocol Document (Authors, 30m) ----------------------------------- Link: http://tools.ietf.org/agenda/81/slides/drinks-0.pptx Syed speaking -- covered the protocol document. (Ken's notes) - At Draft Version 09 - Presented slide listing comments provided since the last IETF. - Presented slides listing updates that were made in response to those comments. - Identified the fact that the Security Considerations section needs improvements in the next rev. - The fact that the current spec allows the SPPP implementer to chose between a stop-and-rollback (all or nothing) policy or a stop-and-commit policy (partial success). This choice will be removed and a stop-and-rollback policy will be required. (Note added Alex: This is also in-line with the requirements document (Sumanth's notes) - He covered the comments received during WGLC (mostly from Dean Willis) and the resolutions; main issues and resolutions: > Security requirements and considerations need more clarification (e.g., authentication v/s authorization) - Some clarifications have been provided > Guidance regarding the transport protocol - Proposed mandatory SOAP-based transport > Questions around IANA registry - Decided against it > Clarifications around the use of Time - Specified the use of UTC > Editorial updates - Completed - Next steps > Clean-up security requirements and considerations - No comments > Clarify the way commands are handled; specifically mandate atomicity of operations - WG Chairs asked if there were any objections; none were noted. - Next Steps > Present an update with the remaining items in the next couple of weeks. > The authors in the room agreed to this, based on a question from Sumanth (as the WG Chair) 4) Transport Document (Author, 10m) ----------------------------------- Ken -- presenting. - acknowledge Manjul for the comments provided. - identified the issues and the changes to the transport document since last IETF meeting - identified the comments from Peter St. Andrae. Acknowledges that these need to be addressed - Next Steps > Present an update with the remaining items in the next couple of weeks > The authors in the room agreed to this, based on a question from Sumanth (as the WG Chair) 5) Open source project (project participants, 10m) -------------------------------------------------- Link: http://tools.ietf.org/agenda/81/slides/drinks-1.pdf Dean providing overview for OpenSPP - mission statement; statement update and other project level updates. - David mentions the need for re-use of the same data model for building other server components, such as, resolution server or distro server - Ken interested in feedback as to whether the existing response codes are sufficient or not - Dean noted that more situations were discovered during the implementation that could use more specific response codes - Dean plans to the bring together developers and protocol contributors to give formal feedback on lessons learned. - Dean extends comment from Jeremy: Validation of messaging data in the database is not sufficient in the spec--(Syed: points to perhaps too much flexibility in the protocol) - Dean notes; that the implementers didn't have to, but they focused on the scalability concerns as well. What was built is in fact scalable to host millions of queries. Gives credits to Jeremy and Vijay for the work. - (David) raised the question around Slide 6. His take was that the tight coupling of data models with SOAP and XSD makes it hard for implementations to separate the data model from the protocol, especially when there are future uses within lookup and distribution. > There were some discussions around this; with Ken clarifying that the data models in the document are not the same that should be used for lookup and - (Jeremy) the draft did not give sufficient data definition for validation of message and database data 6) Discussion around the possible need for a distribution protocol (Draft Authors, 10m) --------------------------------------------------------------------------------------- Link: http://www.ietf.org/id/draft-schwartz-drinks-spdist-00.txt - David is coming to present the SPDP protocol slides. - Slide 1: Is the SPDP needed. (Note: Towards the beginning of the session Gonzalo indicated that the question is open and this work may have to go to DISPATCH) - Use case 1: Local server; SPPP doesn't claim to provide provisioning support from Registry to the Local Server - Use case 2: Bulk data distro from registry to remote local data repo instead of incremental updates, the updates are sent in bulk not triggered by each individual change - Use Case 3: incremental view update, slice of the full set and only the selected slice Switched away to a different slide deck. - Namespace: Local server get update from multiple registry - Incremental data - Optimize bulk distro of data - Notion of versioning - Consistency across repos may be required - Data distrib - pull vs push; sync vs async; what if local repo is unavailable - Data Partitioning: segment the data so it is distributed as a whole across a set of servers - Next steps; DRINKS to address SPDP?; describe the relevant use cases; gain consensus before moving forward - hopes there can be an active dialog on this issue in the list before the next IETF. - mentions that people are asking for use of DNS protocol today since there is no other mechanism available. There is a strong need in this domain for a distribution protocol. - Alex: Notes that there are many different issues identified in the slides; operational + arch - David: Describes how the section of a number space cannot be obtained today in a DNS repl mech which is all or none. - David: commits to putting out a more detailed document before next IETF meeting. - David notes that he has submitted the individual draft and encourages everyone to follow up on it. - David notes that he is using custom solutions to accomplish this today. SPDP is a need in the domain where ENUM is used for resolution. And the first repl solution happens to be IXFR/AXFR, which is the wrong tool - Jon: Advises that he need to clarify the problem space and why the current mechanisms are insufficient - Shockey; agrees to Jon's assessment and notes that a step by step reasoning why DNS repl is broken will go a long way. - When will this be done? - Need volunteers - Standardized data sets over a SOAP UI - (David) will get a few more students IXFR sucks for this purpose (Jon) You need to document this. (Rich) Share this with the DNS - Discuss this on the mailing list - Bring it to DISPATCH in Taipei 7) Next Steps (15m) ------------------- - Shockey; we need to reserve a field within the protocol (David; we did) that can be used for global SPN identification. - Alex; confirms that gSPID work will not happen in DRINKS. - Shockey; agrees. Area Directors wants the work to go on in the DISPATCH and follow up on the DISPATCH list. - Sumanth; repeats the directives to followup on the DISPATCH - Alex; Notes couple of issues in the sppp doc; and that transport doc needs to be fixed as well. - David; confirms that additional work like the REST api will have to be taken to the DISPATCH - Sumanth; confirms; it can be considered new work or DRINKS go dormant while the REST work continues. - Alex: at least 10 people signing up for the work will be needed to keep it open. Or else, the group will have to be closed. - Sumanth; hum on publication of the use case document. ***No objections from attendees*** - Sumanth: Identifies a need for a virtual meeting to completing the protocol work. 5 people showed interest. - Sumanth: Notes that there will be update to the transport document as well. And that there will be a quick last call for both sppp and transport document. - Sumanth: notes updates shall happen within two weeks. - Upcoming work? - The only work so far has been the one suggested by David - Related work (e.g., Global SPID)? (Rich) There is a liaison from ITU that is still valid. (David) We did. (Alex) What are the next steps? (Rich) This should continue on the DISPATCH list (individually submission or AD-sponsored I-D). - WG shutdown before Taipei possible? - Asking for any final comments, discussions, etc. Session closes --------------