IETF
rtgwg
rtgwg@jabber.ietf.org
Monday, July 20, 2015< ^ >
Room Configuration
Room Occupants

GMT+0
[06:56:51] Meetecho joins the room
[06:57:14] Adam Foldes joins the room
[07:01:41] akatlas joins the room
[07:01:43] dhruvdhody joins the room
[07:02:24] akatlas leaves the room
[07:02:45] samante joins the room
[07:03:30] Jeffrey Haas joins the room
[07:04:52] <dhruvdhody> 1) 09:00-09:10 - WG Status Update
Jeff Tantsura, Chris Bowers
10 minutes
[07:05:55] <dhruvdhody> Acee: question on ospf MRT
[07:06:07] <dhruvdhody> and ospf MRT can follow the publication request from RTGWG
[07:06:25] akatlas joins the room
[07:07:22] <dhruvdhody> 2) 09:10-09:20 - draft-liu-rtgwg-ipipv4-tunnel-yang
Helen Chen
10 minutes
[07:08:11] Jeffrey Haas joins the room
[07:08:42] russ joins the room
[07:09:11] Jeffrey Haas leaves the room
[07:10:33] russ leaves the room
[07:10:46] russ joins the room
[07:11:36] Chris Bowers joins the room
[07:15:27] Dhruv Dhody joins the room
[07:15:28] Chris Bowers leaves the room
[07:16:26] Tal Mizrahi joins the room
[07:17:42] Chris Bowers joins the room
[07:18:20] <dhruvdhody> acee on mic
[07:19:23] GIANG NGUYEN joins the room
[07:19:58] Chris Bowers leaves the room
[07:20:03] <dhruvdhody> Concept of bind interface makes an assumption that the interface be a tunnel, they should be parallel.
[07:20:26] <dhruvdhody> Helen: the interface doesn’t have to be physical only
[07:20:36] GIANG NGUYEN leaves the room
[07:21:21] <dhruvdhody> Beniot: in the yang discuss yesterday, another model was discussed where the idea of tunnel type was brought up
[07:21:42] <dhruvdhody> Helen: we don’t want tunnel type to change dynamic
[07:21:49] <dhruvdhody> chairs: take to list
[07:22:00] <dhruvdhody> alia: belong in int-area
[07:22:03] Tal Mizrahi leaves the room
[07:22:30] <akatlas> in int-area WG specifically
[07:22:43] Dhruv Dhody leaves the room
[07:22:54] <dhruvdhody> maybe better to augment the interface, using bind interface is strange
[07:23:03] <dhruvdhody> thanks alia for the correction :)
[07:23:41] <dhruvdhody> Alia on Mic
[07:24:11] <dhruvdhody> IP in IP, GRE belong in INT area. Checked with Ryan and INT area WG is the right place to take this work
[07:24:40] <dhruvdhody> 3) 09:20-09:30 - draft-chen-rtgwg-key-table-yang
Helen Chen
10 minutes
[07:24:54] Robert  Raszuk joins the room
[07:24:56] Adam Foldes leaves the room
[07:25:55] Chris Bowers joins the room
[07:27:21] Tore Anderson joins the room
[07:28:29] Tore Anderson leaves the room
[07:31:08] Chris Bowers leaves the room
[07:33:25] <dhruvdhody> Acee: i am Acee!  
[07:34:08] <dhruvdhody> It would be better if it augments the key-chain model
[07:34:59] Robert  Raszuk leaves the room
[07:35:09] <dhruvdhody> it would be better if we can keep these two separate…
[07:35:30] <dhruvdhody> Helen: Keychain is restricted and is designed for IGP
[07:35:39] <dhruvdhody> Acee: No, it can be attached any place
[07:37:22] <dhruvdhody> Jeff Haas: output from KARP WG, last data store of security information, it was designed for user to interact
[07:37:51] Tore Anderson joins the room
[07:37:53] <dhruvdhody> Question: How are you expecting user/API to interact with this?
[07:37:59] Tore Anderson leaves the room
[07:38:01] <dhruvdhody> is it RW or RO?
[07:38:41] <dhruvdhody> Joel: 7210 was abstract model, WG agree on abstract model
[07:38:57] <dhruvdhody> in terms of KARP, our expectation was some yang model like that
[07:39:34] Robert Raszuk joins the room
[07:39:39] <dhruvdhody> Jeff: it has flexibility , but would API would populate these table, it makes sense split into config and operational state
[07:39:44] <dhruvdhody> Uma on Mic
[07:39:59] <dhruvdhody> what is the conclusion?
[07:40:05] <dhruvdhody> can both coexist>?
[07:40:26] <dhruvdhody> Helen: only one needs to exist
[07:40:38] <dhruvdhody> they can both exist but only one needs to
[07:40:49] Simon Romano joins the room
[07:40:49] <dhruvdhody> Chairs: take to the list
[07:40:57] <dhruvdhody> acee: get ur implementation to the list
[07:41:14] <dhruvdhody> Moving on……..
[07:41:16] <dhruvdhody> 4) 09:30-09:40 - draft-liu-rtgwg-yang-rip
Xufeng Liu
10 minutes
[07:42:09] Robert Raszuk_8072 joins the room
[07:42:25] Robert Raszuk leaves the room
[07:44:57] Russ Mundy joins the room
[07:47:56] <dhruvdhody> Acee: it would be good if it augments (or reference) a keychain
[07:48:08] <dhruvdhody> rather than define our own authentication
[07:48:20] <dhruvdhody> Xufeng: we will do exactly as ospf
[07:48:52] samante leaves the room
[07:49:03] <dhruvdhody> Jeff (BFD Chair): try to use the BFD yang, provide feedback to the WG
[07:49:11] <dhruvdhody> doing this in RIP is very useful
[07:49:30] <dhruvdhody> Acee: your draft still has authentication, perhaps waiting for this discussion
[07:49:53] <dhruvdhody> Acee: this is different from OSPF/ISIS
[07:49:58] <dhruvdhody> need to be checked offline
[07:50:27] <dhruvdhody> Chairs: the work looks complete, after addressing the comments we will take it up for adoption
[07:50:52] <dhruvdhody> 5) 09:40-09:50 - draft-acee-rtg-yang-key-chain
Yingzhen Qu
10 minutes
[07:54:13] Russ Mundy leaves the room
[07:54:52] <dhruvdhody> Acee: we should add attachment point in operational state
[07:55:13] samante joins the room
[07:55:35] <dhruvdhody> there is precedence in ACL model, and they add an opaque string (where the policy is used)
[07:56:09] <dhruvdhody> we don’t have key derivation right now (and can be done via augmentation later)
[07:56:36] <dhruvdhody> can we get a confirmed answer on operational state
[07:57:03] <dhruvdhody> acee: decision cannot be made in this IETF as key members are missing
[07:57:14] russ leaves the room
[07:57:20] Abhay Roy joins the room
[07:58:00] <dhruvdhody> Jeff: regardless of structure, put things in, realize how it would be used, group them correctly, think usecases and application
[07:58:58] <dhruvdhody> Chairs: Joel have ur comments are adressed?
[07:59:11] <dhruvdhody> Joel: Yes (did i hear that correct?)
[07:59:23] <dhruvdhody> 6) 9:50 - 10:00 - draft-shaikh-rtgwg-policy-model
Ina Minei
10 minutes
[07:59:30] Abhay Roy leaves the room
[08:03:42] Russ Mundy joins the room
[08:06:36] <dhruvdhody> Jeff Haas: i have reviewed BGP and this model, this one is mature, and WG should look into it, how this generic works
[08:06:49] <dhruvdhody> Ina: In which WG BGP Policy yang?
[08:07:22] russ joins the room
[08:07:40] <dhruvdhody> Jeff: Here is a good place as any
[08:08:33] <dhruvdhody> stephanie: there is need to define a boundary between IGP and other model
[08:09:14] <dhruvdhody> use of tag is local, you can use tag for any, tag should be generic and good to be extended
[08:09:24] <dhruvdhody> Acee: this is the right model, should be in WG
[08:09:35] <dhruvdhody> good that the split is done between generic and BGP model
[08:10:15] <dhruvdhody> Chairs: Well structured, reasonable
[08:10:26] <dhruvdhody> Who has read?
[08:10:48] <dhruvdhody> Encourage more people to read, this is what operators want!
[08:11:01] <dhruvdhody> 7) 10:00 - 10:10 - draft-li-rtgwg-utunnel-yang
Zhenbin Li
10 minutes
[08:11:11] <dhruvdhody> Robin from Huawei
[08:20:15] <dhruvdhody> Jeff: you are trying to provide a framework for generic, but we have IP tunnel, MPLS TE, how do you want this to work with the other work
[08:20:27] <dhruvdhody> Robin: major use in operational state
[08:20:50] <dhruvdhody> when we try to use tunnel policy or group you refer to this operational data
[08:21:22] <dhruvdhody> Jeff: if this for coordination, there is a bit of work needs to be done here
[08:21:58] <dhruvdhody> Lot of work in tunnel, we have not yet not chosen a thing to come together.. this is an aggressive piece of work.
[08:22:58] <dhruvdhody> Ron : There is laudable effort, but where does it belong INT area or RTGWG, you need to discuss with them.
[08:23:46] <dhruvdhody> You are dealing with routing services, they should be dealt in diff model
[08:24:00] <dhruvdhody> [Name Missed]: How to unify so many tunnels, is it necessary to unify?
[08:24:38] <dhruvdhody> Robin: this is the base, the unified model would not change the config model for each tunnel type
[08:25:59] <dhruvdhody> 8) 10:10 - 11:30 - Routing YANG DT draft-rtgyangdt-rtgwg-device-model
80 minutes
[08:26:51] <dhruvdhody> Lou Berger presenting
[08:40:19] <akatlas> http://www.claise.be/YANGPageMain.html gives the number of YANG models.  http://wiki.tools.ietf.org/area/rtg/trac/wiki/RtgYangCoordSummary has a list of those related to Routing.  (Just FYI)
[08:43:06] <dhruvdhody> And thanks Qin & David for maintaining that wiki :)
[08:43:42] <akatlas> absolutely!  It helps make sense of the forest of YANG models.
[08:44:12] <Jeffrey Haas> +1
[08:48:48] <dhruvdhody> Robin on Mic
[08:49:34] <dhruvdhody> why is VPN used as logical network element?
[08:49:51] <dhruvdhody> Acee: its more than a VRF
[08:50:03] <dhruvdhody> Lou: diff implementation would do things diff
[08:51:02] <dhruvdhody> You have the flexibility to do multiple things
[08:51:10] <dhruvdhody> Dieter in Mic
[08:51:24] <dhruvdhody> is it possible to share the resources between physical and logical device
[08:51:38] <dhruvdhody> Lou: this is hardware implementation, we have not yet discussed at that level
[08:51:53] Simon Romano joins the room
[08:51:55] Simon Romano leaves the room
[08:52:07] akatlas leaves the room
[08:52:09] Jeffrey Haas leaves the room
[08:52:25] samante leaves the room
[08:52:28] samante joins the room
[08:53:17] <dhruvdhody> <not to use an analogy o far >
[08:53:39] <dhruvdhody> for virtual element, do you need to put logical element in data model.
[08:53:49] akatlas joins the room
[08:54:11] <dhruvdhody> For physical device, you have device container and you have to have this logical container even for a very simple device
[08:54:37] <dhruvdhody> its better to have separate netconf session for each logical elements
[08:54:59] <dhruvdhody> the QOS, interface are outside, there is a debeate on whats in and whats out
[08:55:54] <dhruvdhody> Lou: Stack Overflow and crash!
[08:57:28] akatlas joins the room
[08:57:29] akatlas leaves the room
[08:58:07] <dhruvdhody> Acee presenting
[08:58:10] <dhruvdhody> section 2!
[08:59:47] akatlas leaves the room
[09:01:39] Jeffrey Haas joins the room
[09:02:48] <dhruvdhody> lada on mic
[09:03:27] <dhruvdhody> the ietf interfaces was designed to be top level container, do you intent to change the RFC7223
[09:03:36] <dhruvdhody> or change yang to allow mount
[09:03:45] <dhruvdhody> but in yang 1.1 this was abandoned
[09:07:13] dhruvdhody leaves the room
[09:07:25] samante leaves the room
[09:08:46] dhruvdhody joins the room
[09:09:05] Robert Raszuk_8072 leaves the room
[09:09:15] <dhruvdhody> back..
[09:09:34] <dhruvdhody> Chris Bowers on Mic
[09:10:12] <dhruvdhody> 90% use would for default network element and routing instance, it should be easily accessible and made easier
[09:10:31] <dhruvdhody> That would make it user-friendly
[09:10:41] <dhruvdhody> Acee: defer that question till later slide
[09:11:24] <dhruvdhody> Chris: there is complexity in defining multiple ways to get to the same information
[09:12:00] <dhruvdhody> Robin: if we have L3VPN or VSI, are they two different istance? or single?
[09:12:24] <dhruvdhody> Acee: implementation dependent, both are options
[09:12:29] <dhruvdhody> more after the slides
[09:13:11] <dhruvdhody> Acee: ALL WILL BE revealed!
[09:14:02] <Jeffrey Haas> import kitchen-sink { prefix "ietf"; }
[09:14:02] akatlas joins the room
[09:14:29] <dhruvdhody> 😖
[09:15:23] akatlas leaves the room
[09:15:51] <dhruvdhody> Igor: I2RS agent is related to routing element or instance
[09:16:35] <dhruvdhody> Dan: because agent is running on virtual NE, or you can run on device, you get a choice
[09:17:09] <Jeffrey Haas> (for minutes, Dean Bogdanovic)
[09:17:10] <dhruvdhody> Loa: does RSVP has RSVP-Te
[09:17:38] <dhruvdhody> Acee: yes!
[09:18:33] <dhruvdhody> RSVP is individual draft, there are multiple drafts missing but they are individual drafts
[09:18:55] <dhruvdhody> thanks for the correction :)
[09:19:55] dhruvdhody joins the room
[09:20:01] <dhruvdhody> To go back to Robin’s comment… the draft needs to be updated
[09:20:23] akatlas joins the room
[09:20:45] dhruvdhody leaves the room
[09:21:34] akatlas leaves the room
[09:22:12] Jeffrey Haas leaves the room
[09:22:59] <dhruvdhody> Dean Bogdanovic on Mic - NACM can be used for access control list
[09:23:12] samante joins the room
[09:24:03] <dhruvdhody> Chris: How does the hierarchy look in this model
[09:24:17] <dhruvdhody> Lou: in case CE its single element.
[09:24:46] <dhruvdhody> In PE, for each VRF CE would be configured under instance view
[09:25:02] <dhruvdhody> the same hold true to IGP
[09:25:15] <dhruvdhody> this would be a good example to walk through in the draft
[09:25:45] <dhruvdhody> Jeff Haas: There is significant value in the work!
[09:26:03] <dhruvdhody> <<in other words, thank you but get ready>> : Lou
[09:26:57] <dhruvdhody> Jeff: Organizationally this is great but outside of those natural relationships, the rest is quite arbitory like interfaces in top.
[09:29:15] <dhruvdhody> Jeff: Look for yang library work
[09:29:54] <dhruvdhody> Lou: in DT that was discussed, we wanted to have this because we should be able to do this without talking to the device
[09:30:04] <dhruvdhody> as device would not exist
[09:30:14] <dhruvdhody> and allow vendor neutrality
[09:30:53] <dhruvdhody> The programmers (user) feedback was that this is needed and useful
[09:32:05] <dhruvdhody> Beniot (OPS AD): thanks for the work!
[09:32:41] <dhruvdhody> Chair: do we need to talk more in the next session?
[09:32:56] <Russ Mundy> hummmmm
[09:33:06] <dhruvdhody> 15 mins on wednesday session
[09:33:09] <dhruvdhody> more YANG!
[09:33:12] <dhruvdhody> yeah!!
[09:33:33] <dhruvdhody> <END OF SESSION>
[09:34:10] dhruvdhody leaves the room
[09:34:20] samante leaves the room
[09:34:34] russ leaves the room
[09:34:41] Meetecho leaves the room
[09:35:23] Russ Mundy leaves the room
[09:47:41] samante joins the room
[09:59:55] Simon Romano leaves the room
[10:06:41] Simon Romano joins the room
[10:07:06] dhruvdhody joins the room
[10:18:29] dhruvdhody leaves the room
[10:22:14] Simon Romano joins the room
[10:22:25] Simon Romano leaves the room
[10:39:55] Simon Romano leaves the room
[10:50:44] Simon Romano joins the room
[10:51:34] samante leaves the room
[10:53:43] Simon Romano leaves the room
[11:00:54] samante joins the room
[11:03:20] dhruvdhody joins the room
[11:07:53] samante leaves the room
[11:09:34] dhruvdhody leaves the room
[11:12:37] dhruvdhody joins the room
[11:12:42] dhruvdhody leaves the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!