============================================================= NETCONF Data Modeling Language WG (netmod) IETF #81, Quebec City, Canada TUESDAY, July 26, 2011, 13:00-15:00, Room 2101 Minutes Kent Watsen, 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 {Kent Watsen} - jabber scribe {Mehmet Ersue} - blue sheets - agenda bashing - published RFCs: - RFC 6020 (draft-ietf-netmod-yang) - RFC 6021 (draft-ietf-netmod-yang-types) - RFC 6087 (draft-ietf-netmod-yang-usage) - RFC 6110 (draft-ietf-netmod-dsdl-map) - RFC 6244 (draft-ietf-netmod-arch) - status and milestones 2) Core system data model (Andy) [ 20 min ] - draft-bierman-netmod-system-mgmt-00 3) Core interface data model (Martin) [ 30 min ] - draft-ietf-netmod-interfaces-cfg-01 - draft-ietf-netmod-iana-if-type-00 - draft-bjorklund-netmod-ip-cfg-00 4) Core routing data model (Ladislav) [ 30 min ] - draft-ietf-netmod-routing-cfg-00 5) SMIv2 translation to YANG (Juergen) [ 5 min ] - draft-ietf-netmod-smi-yang-01 6) Related work (Martin) [ 15 min ] - draft-bjorklund-netmod-snmp-cfg-01 7) Open mike [ 10 min ] Summary: The NETMOD working group has met for two hours on Tuesday during the 81st IETF meeting. The meeting was attended by 25 people in room with a few additional remote participants. - was discussed and there was consensus in the room to accept this individual submission as a WG document. - and the related IANA document were discussed and these document seem to be close to be complete. An individual submission for basic IP configuration received support in the room to become a WG document. - was discussed and will be revised taking into account the feedback received from routing experts. This document needs more review by vendors/operators/experts in order to make sure the routing model matches reality and sufficiently generic to address operational requirements. - is in WG last call and needs more reviews. Martin Bjorklund's review resulted in a few technical issues that need to be resolved. The document editor is going to bring the issues to the WG mailing list. - was presented but only very few attendees had read the document. The WG chairs will poll the WG (by circulating a proposed charter addition on the WG mailing list) whether there is support to take on this work in the NETMOD working group at this point in time. Actors: - KW = Kent Watsen - DR = Dan Romascanu - JS = Juergen Schoenwaelder - AB = Andy Bierman - MB = Martin Bjorklund - LL = Ladislav Lhotka - DK = David Kessens - PL = Peter Lothberg (operator) - JH = Jeff Haas (operator) - RV = Ruediger Volk (operator) Slides: https://datatracker.ietf.org/meeting/81/materials.html Audio: http://www.ietf.org/audio/ietf81/ Meeting Notes: * Core system data model - AB surveys the system data model, focus is on getting agreement on the scope not on the details (yet) - LL suggests to enclose lists in containers as this makes parsing simpler - JS stated we need to figure out what we leave out and what we keep, e.g. do people really change timeouts/retries frequently? - DR is a bit surprised by the lack of enthusiasm to standardize this. - KW states that complete coverage across devices of different vendors is unlikely to achieve. - PL states that he as an operator likes the work done in this WG and he appreciates the work being done. - JH seconds the operator's interest in having standardized interfaces. - PL perhaps look at this more from a service perspective, e.g. a user signs up for a specific service and it should not matter which box delivers it. - DR wonders whether we should go back to operators (after the road show 8 years ago) to get more input from operators. - JH suggests that a good way to prove things are correct is to write back-ends that generate vendor specific configurations from the data models. In addition, do the other conversion so you can figure out what might be missing from real configurations. - About 6.5 people have read the document, but most of the main contributors have read the document. - There was consensus in the room to adopt this document as a WG document (to be confirmed on the mailing list). * Core interface data model - MB summarized the progress since the last IETF meeting - MB asks whether interface types should be IANA maintained identities rather than an enumeration. - The room was favouring identities, to be confirmed on the mailing list. - There was consensus in the room to adopt the IP interface configuration document as a WG document (to be confirmed on the mailing list). * Core routing data model - LL reviewed the changes since the last version. - PL: How do you present multi-protocol and multi-topology routing protocols like ISIS? - RV: Clarify the terminology of routing processes and the relationship to OS processes. - JS suggests to work with operators to make sure existing real-world routing configurations can be represented in the YANG data model and be converted back. - DR will help ensuring review by routing experts. There has already been one review and another is promised prior to last call. - PL states that the real value is abstraction of details in order to configure the network, not specific boxes. - RV says that he has his own way of representing high-level routing policies that are then translated (often in weird ways) into configurations of boxes; the target should first be on the configuration interfaces provided by vendors but we have to approach the problem bottom up. In particular, the existing configuration is often just an artifact of weird translations given device restrictions and not necessarily a useful basis for study. - PL says that in large networks, you have high-level policy languages that are driving configurations (even though they might not be complete). - RV says for IP configuration, there is not a need to distinguish between IPv4 and IPv6 routing configuration since there is so much commonality. * SMIv2 translation to YANG - JS explains some technical last call comments and he will post them as open issues to the mailing list. - DK asks for more reviews * SNMP configuration data model - MB presented the SNMP configuration data model - Only one person in the room read the document. - AB says its valuable to have this standardized, perhaps going to experimental first and standardize later. - BW asks whether this is high priority? - KW says that he has interest in this work. - Next IESG telechat is in August so there is time to discuss on the list - deadline for this telechat would be August 10th. - KW asks why operators do not use the SNMP interface? - RV says that this was discussed back at an IAB workshop that lead to the creation of NETCONF and YANG. - AB says that also for RMON, people use the CLI and not SNMP. - The chairs will poll the mailing list for support of the work and the proposed charter change.