============================================================= NETCONF Data Modeling Language WG (netmod) IETF #88, Vancouver, Canada Wednesday, November 6th, 2013, 15:50-17:20, Georgia A Minutes Alexander Clemm, Juergen Schoenwaelder ============================================================= WG Chairs: David Kessens Juergen Schoenwaelder WG URL: http://tools.ietf.org/wg/netmod/ Jabber: xmpp:netmod@jabber.ietf.org Agenda: 1) Administrivia (chairs) [ 5 min] - minutes scribe {volunteers welcome in advance!} - jabber scribe {volunteers welcome in advance!} - blue sheets - agenda bashing - status and milestones 2) Core interface data model (Martin) [15 min] - draft-ietf-netmod-iana-if-type-07 (WG last call completed?) - draft-ietf-netmod-interfaces-cfg-12 (WG last call completed?) - draft-ietf-netmod-ip-cfg-11 (WG last call completed) 3) Core routing data model (Martin, Ladislav) [15 min] - draft-ietf-netmod-iana-afn-safi-00 (ready for WG last call?) - draft-ietf-netmod-routing-cfg-11 (ready for WG last call?) 4) Core system data model (Andy) [ 0 min] - draft-ietf-netmod-iana-timezones-00 (WG last call completed) - draft-ietf-netmod-system-mgmt-08 (WG last call completed) 5) SNMP configuration data model (Martin, Juergen) [ 5 min] - draft-ietf-netmod-snmp-cfg-03 (ready for WG last call?) 6) Proposals for new infrastructure work [25 min] - draft-bjorklund-netmod-yang-xpath-extensions-00 [ 5 min] - draft-bierman-netmod-yang-conformance-01 [ 5 min] - draft-lhotka-netmod-yang-json-02 [ 5 min] - draft-clemm-netmod-mount-01 (IPR) [ 5 min] - discussion / prioritization [ 5 min] 7) Proposals for new data models [15 min] - draft-clemm-netmod-yang-network-topo-01 [ 5 min] - draft-huang-netmod-acl-03 [ 5 min] - discussion / prioritization [ 5 min] 8) Rechartering discussion [10 min] Additional related documents: * http://tools.ietf.org/html/draft-yeung-netmod-ospf-00 * http://tools.ietf.org/html/draft-zhdankin-netmod-bgp-cfg-00 * http://tools.ietf.org/html/draft-zhou-netmod-openv6-transition-cfg-00 Summary: About 60 people attended the NETMOD meeting in Vancouver. The chartered documents and the associated actions items are listed below: - draft-ietf-netmod-ip-cfg has been sent to the AD (Benoit). - draft-ietf-netmod-interfaces-cfg and draft-ietf-netmod-iana-if-type will both be revised once more to change the interface type to an identity. The documents will then be sent to the AD (Benoit). - draft-ietf-netmod-routing-cfg will be updated after the IETF meeting and then put into WG last call. - draft-ietf-netmod-system-mgmt and draft-ietf-netmod-iana-timezones have completed WG last call. The shepherd (Juergen) has to produce a writeup and then the document will be send to the AD (Benoit). - draft-ietf-netmod-snmp-cfg has been updated shortly before the meeting and the authors believe it is complete. It will go to WG last call. The remaining meeting time was used to look at some of the proposed new work. The meeting went short of time and further discussions of proposed new work and whether the WG should recharter or close down once all chartered documents have been delivered will take place on the mailing list. Actors: - JS = Juergen Schoenwaelder - AB = Andy Bierman - MB = Martin Bjorklund - LL = Ladislav Lhotka - DK = David Kessens - BC = Benoit Claise - KW = Kent Watsen - PS = Phil Shafer - AC = Alexander Clemm - EV = Eric Voit - BL = Balazs Lengyel - JM = Jan Medved - CM = Carl Moberg - L? = Lisa ? Slides: http://www.ietf.org/proceedings/88/netmod.html Audio: http://www.ietf.org/audio/ietf88/ Meeting Notes: * Core interfaces data model MB presented a proposal that was also circulated on the mailing list to use identities for interface types. The people in the room who were involved in the discussions and who have read the document think this is a good way forward. Martin will post revised documents and then this will be quickly confirmed on the mailing list and then the documents will be with BC. * Core routing data model LL reviewed the changes made in the routing model document. MB: I proposed to use a name instead of a numeric identifier. MB: Can the numbers change during reboots? LL: Probably not a good idea to change the identifiers for routing instances and system controlled RIBs. LL: I think it is acceptable to use names as keys instead of numbers. PS: Does the system create these routing instances or are they configured? LL: The system creates them. PS: Are the number fixed or can they change? LL: It might be problematic if they change. PS: Can I2RS not use the name as well? LL: They insist on using numerical identifiers and they want to have they identifiers numeric. AC: We need to align our I-Ds containing BGP and OSPF data models with the core routing data model. LL will revise the document and then it will get last called. * Core system data model JS to produce a writeup and to send the document to the AD (Benoit). * SNMP configuration data model MB explains that text has been added discussing co-existence issues (stemming from different persistence models of SNMP and NETCONF) that implementers should consider. This document will go to WG last call and a high threshold will be applied to ensure this document gets proper review. * Proposals for new infrastructure work MB presents a proposal for a YANG extension that can be used to define XPATH functions to be used in when and must expressions. LL: I support this work but I believe this requires to bump the YANG version number. PS: I believe this should not require a YANG version number change. MB: The change affects a YANG compiler that parses XPATH expressions. PS: But it does not affect NETCONF server implementations. LL: The extension statements have side effects on other YANG statements. MB: This is why this document will update RFC 6020. AB: This is somewhat a gray area since the XPATH expression remains to be a valid XPATH expression. AB: We need to think how we extend YANG. The IEEE way of doing a number of extensions that are merged at a certain time into the baseline may be a good approach. AB presents a mechanism to define conformance profiles for YANG data models. EV: Is the idea to standardize packages? AB: Good question, could be both. MB: This seems similar to SMIv2 module conformance, not to agent capabilities. MB: Can I create a package and claim conformance to a module? AB: No, you would claim conformance to your own package. PS: I see the value of this from the management application point of view. If a server supports a certain package, a management application can manage it. AB: No, it is a contract in the middle. BC: How does this I-D solve the import by revision problem? AB: Because you have flexibility with the min-revision and max-revision statements. Details are in the I-D. LL provides an update of his YANG to JSON mapping. The mapping is already used by RESTCONF. AC presented a proposal to mount remote datastores into a NETCONF server. JS: Why do you believe this belongs to NETMOD and not NETCONF? AC: The core of the proposal is the YANG module. JS: Can you run an edit-config on mounted data and if so how do you make it atomic? AC: The reading use case is relatively trivial, writing is of course more difficult. BL: I believe the protocol issues need to be address for this to be useful. JM: There are a number of operations that do work without special protocol work, but of course, edit-config would require some transaction support. CM: We have to be careful about the semantics in YANG and how they apply to distributed datastores. JM: The use case is that you have to compile the device model into the controller. CM: I agree and I am talking from experience. * Proposals for new data models Skipped due to a shortage of time. * Discussion and announcement DK: Do we need a new charter or do we simply close down and declare success? AB: Data models like topology data models or packet filter data models need to go for a WG forming BOF, they should not be done here. L?: We need to get more WG people involved in reviewing. People propose new work but chartered work is lacking quite a bit behind. DK announces that he will step down as a co-chair. People should talk to BC if they are interested in becoming a co-chair. DK will continue his duties until a new co-chair has been established.