==================== Ops Area Minutes ============== Taken by Tianran ------------------ 1. Overview of ops-related/important BOFs. - QUIC and PLUS (in TSV area) - What do they have to do with OPS? We need to look at this from our OPS tooling (operation and management) aspects: - Will these new "transports" invalidate some of our tools? - Will firewalts etc still work properly? - Will we be able to adapts our tools to work with these new "transports" So there may be problems or opprotunity for impact in our OPS area So please pay attention - Impact on ACLs etc. 2. Joel on where he sees non-management work going in the next 2 years - sometimes feels as perpetual, but I have seen good productivity and we finished some work we also closed WGs who were not meeting objectives we started some new work work does evolve; things get deployed - interdisciplenary: operational management; other ADs. - Flexibility: - reorganize around changing demands - in future: possibly have new OPS groups chartered with non-OPS AD - exisiting WGs may shuffle into OPS via WG chartering (sidrops) - operations is gonna be different. - v6 transition technologies don't have bright future - shedding legacy technologies - network operators are orchestrators of complex systems - software tooling evolution runs faster than customer -> vendor interaction - yang model catalogs is one example of this. other proprietary/enterprise/open-source interactions - traditional IETF work seems to take too much time, so that is why the above catalogs started for example. - So OPS has to stack up Q&A - Dan York some times it is not clear where to bring back deployment experience. Good too see this is recognized - Dean Boganovich Yang catalogs. Sometimes people do not want to implemnet deploy until it is an RFC We should be flexible Joel: Other SDOs look at us, They are dependent on our progress. Needs to be a concerted effort that requires enery to achive. Sometimes: lack of stability is harmfull Benoit: We saw it coming, we are doing tools in hackaton. But I got 1000 yang models. We have some roots of models that are KEY. We should target the key modules and focus to get those RIGHT. Dean: it is even getting to 2 years. There are too many small issues that come up. So I want flexibility: Maybe an interim-rfc or some such, so that people can work on this Bert: is "implementation agreements (based on a snap shot of draft(s) o possibility? Not sure it is a good idea, but maybe it motivates vebdors to implement Joel: Versioning is getting better and maybe also helps Benoit: If customers tell a vendor: I want version x of a yang datamodel, that may help too. Some want a specific RFC, otherse mayt be heappy with the more flexible versioning scheme of modules Joel: Sometimes demands from operators are not real either, but only to try an motivate implementors. Not sure that is/was good But we have a much more mature space these days But there are challenges to get stuff into products. End of meeting. ============== OpsAWG Minutes ============= Taken by Joel. Agenda bashing Documents TACACS+ protocol rev +wg last call first preso mud - Eliot Lear -------------------------------- referral to slides from item 95 for flow description how should we do extensibility no capabilities exchange or negotiation security considerations need review / consider attestation requesting adoption Dan Romscanu - supportive what do you want adopted Eliot draft lear item netted mud + dns which will be included Henk b Fraunhoffer likes it how to squeeze self descriptive information for composite devices cobra - might be more extensible Eliot - masa server pointer to Anima group Warren - do people like the idea? not an adoption call * hum in favor none really opposed strong support - will call for adoption when new draft arrives Next presentation service models explained - Qin Wu speaking Mehemet Ertsu - narrow down the effort - disagree with term service Benoit Claise spending to much time in my life trying to explain this (this is an important topic) Ji Wu - what is the intention of draft Qin define what a service model is and scope Adrian Farrel -------------------------------- Requirements of Composed VPN service model Hui Deng Michael scharf nokia - Do we understand this as a proposal for a yang model? What’s the use case for having everything in one module? Bert Wijnen - are segments within one operator? Dan Bogdanovitic - how many operators are participating? Benoit - Need other operator input ? Gonzalo - not 100% happy -------------------------------- next preso - automatic attachment of end stations and network devices - Dan Romscanu presenting Paul - Eliot lear - don’t have any objection ? nokia - Dan Romscau - goes a little further than the bootstrap process work -------------------------------- next presentation - Export Community information in IP flow information export - rong gu Benoit - Comments ipfix registry is expert review yes this BGP community is important possibly we opened pandoras box Warren - wasn’t involved in ipfix - how many here were (some) seems reasonable * Asking for a hum for interesting to work on? Some in favor none opposed. -------------------------------- next topic Inband OAM - Frank Brockners key question - here does it deploy? keep inso Warren - what happens when you add stuff? (e.g. mtu issues) Dan Bogdanvitch - our existing troubleshooting tools are not well suited to virtual environments Chris Morrow - wary of anything that steals cpu cycles on routers Dan Bogdanovich - tested on broadcom cpu - trident2 Alt Morton - this thing really is hybrid in the context of active and passive Benoit - is there interest in working on this topic- do we think we should work on this * hum, some in favor. not opposed --------------------------------- Jabber Log: ---- joel jaeggli 9:59 IETF96 - joint opsawg opsarea chat 10:01 adrianfarrel [adrianfarrel@jabber.org/1962c9745b4ff44f] entered the room. 10:01 Yuji Tochio [1801907584@tiergarten.conf.meetecho.com/MeetechoWebLite] entered the room. 10:02 George Tsolis left the room. 10:04 bclaise [bclaise@cisco.com/b7af256a-a06e-4988-984b-270801d65f00] entered the room. 10:04 DanYork [dyork@jabber.org/dhcp-b32a] entered the room. 10:04 DanYork has set the topic to: IETF 96 DanYork 10:04 Hello... I'm jabber scribing. 10:04 Do we have anyone remote? Joe Clarke 10:04 Yes. 10:05 George Tsolis [_1010987355@tiergarten.conf.meetecho.com/MeetechoWebLite] entered the room. 10:05 Joe Clarke_9268 [2100347893@tiergarten.conf.meetecho.com/MeetechoWebLite] entered the room. 10:06 Joe Clarke left the room. DanYork 10:06 Eliot Lear is speaking now about Manufacturer Usage Descriptions 10:07 Joe Clarke [_1347260065@tiergarten.conf.meetecho.com/MeetechoWebLite] entered the room. 10:07 bclaise left the room. 10:08 bclaise [bclaise@cisco.com/8243584f-e2c7-41d2-b484-4e2e209fea1a] entered the room. 10:09 Joe Clarke_9268 left the room. DanYork 10:09 If you have any questions you would like relayed to the mic, please preface it with "MIC:" (or "mic:") 10:14 Eliot asks: How should we do extensibility? 10:17 Laurent Ciavaglia [1867602187@tiergarten.conf.meetecho.com/MeetechoWebLite] entered the room. 10:18 bclaise left the room. 10:19 bclaise [bclaise@cisco.com/317a6393-83df-49ae-9f36-7714d6fa5960] entered the room. 10:19 bclaise left the room. DanYork 10:19 Dan Romascanu at mic 10:19 Laurent Ciavaglia left the room. 10:20 bclaise [bclaise@cisco.com/296e5e88-8a86-460b-9c9d-604b859df916] entered the room. DanYork 10:20 (?) at mic (missed his name) 10:20 Laurent Ciavaglia [_252649981@tiergarten.conf.meetecho.com/MeetechoWebLite] entered the room. 10:22 Evgeniy Zhukov [374561402@tiergarten.conf.meetecho.com/MeetechoWebLite] entered the room. DanYork 10:24 It was Henk Birkholz at the mic Joe Clarke 10:24 hummmmm DanYork 10:27 Next is "Service Models Explained" 10:27 Presented by Qin Wu DanYork 10:36 (Several people at mic) 10:36 Adrian Farrell at mic 10:36 Hui Deng now presenting on "Requirements of Composed VPN Service Model" 10:41 Evgeniy Zhukov left the room. DanYork 10:43 Michael ? from Nokia at mic 10:46 Bert at mic 10:47 Dan Adamovich (?) at mic 10:47 (I'm not sitting near the microphone and so I can't get everyone's names 10:48 Bert at mic 10:49 (?) at mic 10:49 Benoit at mic 10:50 Next presenter - Dan Romascanu presenting on "Automatic attachment of end stations and network devices" DanYork 11:00 (someone at mic... didn't say his name) 11:00 Eliot Lear at mic 11:01 (?) at mic (couldn't understand name) 11:02 Next up - Rong Gu about exporting BGP info in IP Flow Information Export (IPFIX) 11:06 Shishio Tsuchiya [2004471546@tiergarten.conf.meetecho.com/MeetechoWebLite] entered the room. DanYork 11:08 Benoit at mic 11:26 Evgeniy Zhukov [374561402@tiergarten.conf.meetecho.com/MeetechoWebLite] entered the room. 11:27 adrianfarrel left the room. DanYork 11:29 There's a line at the mic. Bert was last at the mic. Then Dan (?) at mic. Then both 11:29 Chris Morrow at mic 11:33 AHA! It is "Dean Bogdanovic" who has been at the mic earlier! (When I've been saying "Dan (?)" or "Dan Adamovich") 11:33 Al Morton at mc 11:33 s/mc/mic/ 11:35 Benoit Claise at mic Joe Clarke 11:35 hummmmm DanYork 11:36 Now switching to "Ops Area". DanYork 11:37 Benoit and Joel Jaeggli coming to the front of the room DanYork 11:38 Only two agenda items: 1) discussion of QUIC/PLUS BOF; 2) Joel pontificating on where he sees things going. DanYork 11:45 Dan York (me) at mic, just emphasizing the point that QUIC is not a theoretical protocol in development but rather a protocol already in use today by Google (and a few others) 11:45 Eliot Lear at mic 11:49 You have disconnected 12:02 You have connected DanYork 11:59 Dan York (me) back at the mic 12:00 Dean Bogdanovic at mic 12:02 DanYork has set the topic to: IETF 96 12:02 George Tsolis [_1010987355@tiergarten.conf.meetecho.com/MeetechoWebLite] entered the room. 12:02 DanYork [dyork@jabber.org/dhcp-b32a] entered the room. 12:02 Shishio Tsuchiya [2004471546@tiergarten.conf.meetecho.com/MeetechoWebLite] entered the room. DanYork 12:06 Bert at mic 12:12 Joe Clarke left the room. DanYork 12:12 And we're done...