IETF 112- LISP WG Minutes CHAIR(s): Joel Halpern ( jmh AT joelhalpern.com ) Luigi Iannone ( ggx AT gigix.net ) SECRETARY: Padma Pillay-Esnault ( padma.ietf AT gmail.com ) AGENDA Session 1/1 (120 Minutes) =-=-=-=-=-=-=-=-=- Friday, November 12, 2021 12:00 - 14:00 (UTC) Administration Halpern/Iannone - Agenda Bashing - Status reports for WG drafts 10 Minutes (Cumulative Time: 10 Minutes) Luigi: Since 101 some documents have been sitting there for a while and the bottleneck is LISP-Sec. Work on LISP-sec, pen passed to Damien. Aim to finish this doc before Christmas. As for other docs, Yang model is ready. Today, we will go over the LISP VPN and Lisp L2/L3 EID Mobility Using a Unified Control Plane and discuss them. And there are a bunch of doc sitting for a while and need some attention to gauge interest in the WG. Slides are found here https://datatracker.ietf.org/meeting/112/materials/ o WG Items - LISP L2/L3 EID Mobility Using a Unified Control Plane https://datatracker.ietf.org/doc/draft-ietf-lisp-eid-mobility/ 15 Minutes (Cumulative Time: 25 Minutes) Marc Portoles Discussions: Marc concluded the slides with the following question: whether site-id should be used instead of ES-ID? Marc mentioned this had been used quite a lot and it is working in practice. Luigi Ð You are using Instance-ID to distinguish between L2 and L3 and you do not distinguish these ID. There is no reservation except saying this ID is used for L2 by config? Marc: Yes. Exactly Ð by defining the instance-id we map it to a vrf/ vlan and link it. Luigi Ð How about LISP GPE? Can we use it here? Marc Ð Yes. There is a section for encapsulation and we can combine L2/L3 and segment path as well? Luigi Ð It would make sense to me to reuse the site ID rather that the new segment ID. You should revise the document to reference to the documents we have now and look if there is any implications in LISP-Sec. Marc Ð Sure. - LISP Virtual Private Networks https://datatracker.ietf.org/doc/draft-ietf-lisp-vpn/ 10 Minutes (Cumulative Time: 35 Minutes) Marc Portoles Discussions: Authors: would like to request WG Last Call. Luigi: Polled for any questions from audience? Luigi: You mention that AFI is a distinguished name however in the doc there is a distinguished names type that is not defined anywhere. Marc Ð we use the type 17. Joel: you're going to use distinguished names then you're going to need to put in text about how they are distinguished? What keeps them separate because your example is just an arbitrary name that seems to imply that they're going to collide in the mapping system. That's not good but if you can keep them separate then specify how. Marc: Okay Luigi: Should also start by putting the document in the reference. Generate update needed for the doc to fix the references. Another question: You show how it goes in one direction but how will it go also in the other direction if there is any mismatch? Was this discussed? Marc: Valid point the document does not discuss this. Luigi: Similarly, when you have several xtrs and then you want to update the mapping you need to add text Joel Ð I went to see at the draft and I see you imply that the names have semantics and the names do not have semantics. Luigi Ð in the light of the comments here, suggest that first we do a revision of the documents before we move forward. Marc - OK o Non WG Items - LISP Map Server Reliable Transport https://datatracker.ietf.org/doc/draft-kouvelas-lisp-map-server-reliable-transport/ 20 Minutes (Cumulative Time: 55 Minutes) Marc Portoles Discussion: Authors request to be a WG doc. Luigi polled if there is any questions from the audience. Luigi: There is a registration using UDP but it is not clear how you set up a session. Is there an explicit signaling. Marc: What we do today, there is a try and fail on different transport and then UDP. Luigi: so you send some messages and see if the other side is responding. You can update the document to describe better how these transition occur and how to implement it. Luigi: Personal observation, you authenticate but you do not have security layer. If this document moves forward, the security review may well come back to say this system is not secure. Suggest to have a look at how LISP-Sec apply. Just authenticating to the other side is not enough. Need to update the document Luigi: Regarding adoption for now the draft is expired. Need to do few updates for it to be considered. Like the document for reliable transport it is interesting. As chair I shared my concerns and priority is LISP-Sec for all the documents. We have done a lot of work and this is the last piece that is missing - LISP - Fix 20 Minutes (Cumulative Time: 75 Minutes) Sharon Barkai IPFIX shift from traditional use, seen as useful for sampling in networking. Authors appreciate every feedback and offline very welcome. Discussion: Padma: Very interesting study Ð question regarding sampling, you mention that you are looking at reduction of amount of sampling. My questions is regarding the priority of those packets and whether this large amount of traffic seen as control traffic or just regular traffic for the apps? Sharon Ð using IP fix and NetFlow of cisco, using IP headers Padma: Will take offline with you interested in understanding how the competition between that amount of sampling priority may impact control traffic. Sharon: I think the question is after the sampling has been gathered per application the sampling records are the high priority traffic just to protect the process of ongoing sampling so analysis will be you know protected. You think we can answer that offline and that's a great point thank you. Marc: Could you give an insight of how are you structuring the EIDs or how do you see it being structured Sharon: Great question so currently you know it's a private space of cyber network and it relates to your presentation also Marc and there's about space conservation. We match a specific ID to any group testing or do we simply algorithmically generate EIDs and use the mapping system for a longest match for the best collector for this LISP Nexagon Parking Detection deployment Discussion Luigi: These are interesting use cases for LISP. Do you think there will be extension of LISP Protocol specification? Sharon: May be but so far is enough. Perhaps on how we allocate EIDs, signal partitions.. but so far specification is good enough. Marc: in practice um with signal free and replication, have you seen any scalability issues when deploying this? For example if you have to replicate to too many sites? Sharon: Good point great point so actually for these networks where there's a lot of production of data routed to somewhere close for reduction and then propagation based on a subscription. Marc: Are you doing priority here or are you just doing a selection of list compiled from signals free registration. Then you get all these car logs in the list and then what you do is filter down. Is there are issues with the latency due to distance. Sharon: It doesn't relate to where the car is because that even though the car is moving IP anchor is not moving. ItÕs mostly related to if I put an H3 EID service there then the car can upload more data per meter so IÕll have more coverage and I can get more out of any car so now I have to put the eids based on load of the current compute situation and SLA of the carrier.