What: Combined OpsAWG / OpsArea When: March 28, 2019 16:10 - 18:10 Thursday Afternoon session II Where: Congress Hall 2 OpsAWG Section -------------- Minutes taker: Joe Clarke with transcription from recording Jabber Scribe: Warren Kumari Administrivia - scribes, minutes, current draft status, etc. Tianran / Joe 10 minutes Ignas: questioned whether or not rev -13 of the TACACS+ draft is forthcoming or if it had already been posted Joe: rev -13 had already been posted Network Telemetry Framework Haoyu Song Draft: https://datatracker.ietf.org/doc/draft-song-opsawg-ntf/ 10 minutes Joe: Feels this draft needs to get more comments and feedback from other SDOs (collaboration) [chair hat] Joe (as contributor): I would like to see more blending of telemetry to help provide more advice on value Huawei contributor (didn't catch the name): Communicating with China Standard Association and CHINOG and other NOGs to collector more operator feedback; will come back with that to opsawg Joe: Please all review this document as it is a working group document Secure Device Install Warren Kumari Draft: https://datatracker.ietf.org/doc/draft-wkumari-opsawg-sdi/ 10 minutes Few people have read the draft Tim Carey (Nokia): why does this need standardization Warren: it doesn't per se, but I want vendors to do it Tim Carey: is this informational Warren: yes, this is informational Tim Carey: seems reasonable; maybe BCP Tim Carey: What hardware have you tested this on? Warren: Linux workstation and an OpenWRT router Bill Fenner: Arista looking at what we can do in this direction; will take it back to see if we can do it Warren: Woot Joe: What is your intention with this document? Warren: Read it and provide feedback; maybe adopt it if worthwhile Bill Fenner: What do you think about SSL and authenticating the network server? Warren: From anywhere you can get the config file, this is fair game Warren: But SSL will need a trust anchor Laurent Ciavaglia: ANIMA has done some analysis, which is why they came up with their more complex solution Laurent Ciavaglia: Are you not oversimplifying? Warren: ANIMA is doing a much larger thing with a new control plane; this is simpler as it's focused on a specific use case Warren: An attacker could spoof the network server and serve a bad config file Warren: In my use case, this would be a POP or IX where the bigger worry is smart hands stealing my config Luis Contreras (Telefonica): Test that equipment is stolen and used by an unauthorized party Warren: Maybe; probably needs 802.1AR Eliot Lear had a question on keying that will be taken to the list for purposes of time. Yang data model for Terminal Access Controller Access Control System Plus Bo Wu Draft: https://datatracker.ietf.org/doc/draft-zheng-opsawg-tacacs-yang/ 10 minutes Joe (as chair): ietf-aaa should be a separate module Joe (as chair): Work seems consistent with opsawg; good that it's augmenting the system container in ietf-system Document has been read by a fair number of people in the room Joe (as chair): Please review this document and send to the list; more feedback and restructuring is needed before adoption A YANG Data Model for SD-WAN VPN Service Delivery Bo Wu Draft: https://datatracker.ietf.org/doc/draft-sun-opsawg-sdwan-service-model/ 10 minutes Charles Eckel: Thinks there is value in this draft, especially with the alignment to ME Works with both SDOs More work to do in MEF Terminology well-aligned Tim Carey: Is both IETF and MEF building models? Charles: There has been some work, but the consensus in MEF is that IETF is the right place for this Joe (as chair): More feedback from the list is required Tianran: A similar SD-WAN model has been posted in RTWG; what is the difference? Bo Wu: This work is focused on the SP providing a SD-WAN service only (only one domain), ONUG wants to focus on inter-domain SD-WAN ONUG feels that RTWG is more appropriate for their work Doug Montgomery: Is there an interlock between IETF and MEF? Bo Wu: Nothing formal has been established. This is mostly opportunistic with interested parties trying to do the right thing. Doug Montgomery: Seems kind of opportunistic Charles Eckel: Nothing formal yet; right now it's people working together across SDOs to do the right thing; trying to be more agile with less overhead at the moment; people wants to see this done within the IETF Ignas: Situation described well as to why this ended up in different WGs Ignas: One concern: multiple interested groups building models around one common area; will those solutions be compatible? Ignas: no fundamental difference between SP and Enterprise; trying to separate so strictly may not be the right answer Ignas: For now, since the technology domain is in the IETF, that's why the work is being done here Charles Eckel: On the SP side, it's from the point of view of the customer to the SP; customer will not do anything with the underlay; high-level interface to the customer Charles Eckel: In ONUG, the customer operates the whole service (underlay and overlay); that model will have to be much more detailed; MEF will be much simpler YANG Data Model for Composed VPN Service Delivery Roni Even Draft: https://datatracker.ietf.org/doc/draft-evenwu-opsawg-yang-composed-vpn/ 15 minutes A few people have read this draft Joe: What are your next steps? Roni Even: We want to adopt this draft and move it forward Joe: More comments are needed, especially from operators Joe: Who think this work is valuable? (Less hands than read it) Tianran: How does this module compare to the modules discussed in bess? Roni Even: This is a composed VPN that is composed from segment VPN Qin Wu: This is describing the services model vs. the network model Network Slice Provision Models Shunsuke Homma Draft: https://tools.ietf.org/html/draft-homma-slice-provision-models-00 15 minutes Joe: Where else are you considering presenting this? Shunsuke Homma: Not wanting to create a different definition Liang Geng (China Mobile): This has been discussed previously in Prague and London IETF meetings and in the COMS BoF; it was decided work would not proceed Shunsuke Homma: This work clarifies a concrete definition for slicing alone Tianran: What is meant by models Laurent Ciavaglia: Provisioning models; will be presented in routing Friday; this is akin to the service models draft, which is why it may fit in opsawg Liang Geng: Would like this work to be done in IETF Ignas: BoF conclusion was that net slicing (overall) is not a problem for the IETF to solve; but partial problems are being worked on in routing Ignas: Orchestrating parts of NS is likely the right work for the IETF to do Liang Geng: Would like to see work done on this even if it's not called "slicing" per se Joe: This needs more review and likely more of the next steps (i.e., the model developed) to see this work with in opsawg Joe: I'd like to see more reviews from the routing area on this; get domain expertise to ensure the models are defined correctly Tianran: I agree; more comments on list needed to move this work forward Framework for Automating Service and Network Management with YANG Qin Wu Draft: https://www.ietf.org/id/draft-wu-model-driven-management-virtualization-02.txt 15 minutes Robin Li: Should include list of IETF modules within the draft Robin Li: There should be more coordination, perhaps in GitHub, for module development Robin Li: Will this only be presented in opsawg? Qin Wu: This was presented in RTGWG Joe had to cut the mic; more conversation will have to go to the list Network-wide Protocol Monitoring (NPM): Use Cases Yunan Gu Draft: https://datatracker.ietf.org/doc/draft-chen-npm-use-cases/ 10 minutes No time for comments at mic CCMIB Sean Turner 5 minutes Sean wants no one to comment and no one commented Sean's work will be AD-sponsored Ops-Area Section --------------------- Administrivia - scribes, minutes, etc. Warren / Ignas 10 minutes No comments at the mic