IETF
bfd
bfd@jabber.ietf.org
Monday, July 21, 2014< ^ >
Room Configuration
Room Occupants

GMT+0
[00:39:58] noboakiya joins the room
[00:40:09] noboakiya leaves the room
[19:02:21] adrianfarrel joins the room
[19:05:16] adrianfarrel leaves the room
[19:16:03] adrianfarrel joins the room
[19:19:19] noboakiya joins the room
[19:21:07] kiranmak joins the room
[19:21:56] Jared Mauch joins the room
[19:22:10] <Jared Mauch> Status Slides
[19:22:27] <noboakiya> Starting.  Administrivia: http://www.ietf.org/proceedings/90/slides/slides-90-bfd-7.pptx
[19:22:53] <Jared Mauch> MIB RFC should be shipping soon
[19:23:25] <Jared Mauch> Jeff: Should there be a YANG model to compliment MIB model or supplant it?
[19:23:48] <Jared Mauch> Please review BFD Seamless draft
[19:24:54] <Jared Mauch> *i am scribing, will present any questions at the mic*
[19:26:44] <Jared Mauch> BFD Stability draft being presented now
[19:27:03] <Jared Mauch> (by Ashesh Mishra)
[19:28:33] <Jared Mauch> no stability of BFD engine itself (software or hardware)
[19:28:57] <Jared Mauch> http://tools.ietf.org/html/draft-ashesh-bfd-stability-00
[19:29:44] <Jared Mauch> if timestamp is added at start/end of engine you can measure the engine
[19:32:46] <Jared Mauch> open for questions
[19:33:18] <Jared Mauch> at mic: adding timestamps breaks hardware support of BFD
[19:33:28] <Jared Mauch> (stewart bryant)
[19:34:29] <Jared Mauch> when in development: adding timestamps was useful to diagnose trouble in ecosystem
[19:35:14] <Jared Mauch> stewart point #2 : this is taking the role of loss/delay systems, taking us in the direction of y1731 - all OAM in one
[19:36:57] <Jared Mauch> at mic: this crosses the line into active measurement
[19:38:41] <Jared Mauch> (at mic) greg mirsky: what are we measuring?
[19:40:46] adrianfarrel leaves the room
[19:41:07] <Jared Mauch> shariam: when every packet changes authentication is difficult
[19:41:59] adrianfarrel joins the room
[19:42:30] <Jared Mauch> at mic: (mahesh) we had difficultly troubleshooting why only bfd packets were lost and see value here
[19:43:42] <Jared Mauch> http://tools.ietf.org/html/draft-ietf-bfd-seamless-use-case-00
[19:43:46] <Jared Mauch> Sam Aldrin presenting
[19:45:17] <Jared Mauch> with all SDN/NFV - new architectures and solutions - bfd can be used to measure the forwarding plane if the control plane is not within the device ..
[19:45:36] <Jared Mauch> want to adapt bfd rates based on network usage
[19:46:50] malleus.errata joins the room
[19:47:15] <Jared Mauch> want to use BFD to measure virtual networks and machine connectivity.
[19:47:50] <Jared Mauch> S-BFD, semaless, segment, single, simple, some, sexy ?
[19:48:36] <Jared Mauch> questions?
[19:49:31] <Jared Mauch> Next presenter.
[19:50:00] <Jared Mauch> S-BFD update.. Why another flavor of BFD?
[19:50:16] <Jared Mauch> Presenter: Nobo Akiya
[19:51:40] <Jared Mauch> currently focused on use case, base, isis advertisement
[19:51:51] <Jared Mauch> next focus: SPRING (segment routing dataplane)
[19:52:06] <Jared Mauch> do we need SFC?
[19:53:29] <Jared Mauch> when there's no IP on the device, will be special use case, no udp replaced by ACh type
[19:54:01] <Jared Mauch> if you ave opinion of UDP vs other, please share on-list.
[19:54:32] <Jared Mauch> S-BFD terminology slide, defined in sbfd-base-01
[19:54:47] <Jared Mauch> need input: Option #1 - currently defined in base
[19:55:06] <Jared Mauch> option #2 - alternative, reflects received state or sets ADMINDOWN
[19:56:18] <Jared Mauch> option #1 can't reuse FSM from rfc5880
[19:58:16] adrianfarrel leaves the room
[19:59:16] <Jared Mauch> Seeking input #2 - discriminator uniqueness within admin domain
[19:59:43] <Jared Mauch> is relying on configuration ok, or some other technique?
[19:59:53] <Jared Mauch> seeking input #3
[20:00:10] <Jared Mauch> should some sets of discriminators be reserved?
[20:00:54] adrianfarrel joins the room
[20:01:20] <Jared Mauch> any final questions?
[20:01:43] <Jared Mauch> mic: want to reuse the same version # or new?
[20:02:34] <Jared Mauch> don't need to rev number, but could...
[20:04:18] <Jared Mauch> next preso: Extended BFD discriminator TLV for MPLS ping
[20:04:39] <Jared Mauch> Presenter: Prasad Govindan
[20:05:36] <Jared Mauch> rfc5884 describes using MPLS echo to bootstrap BFD.  There is a need to bootstrap multiple BFD sessions per MPLS FEC LSP, described in document
[20:07:35] <Jared Mauch> another open item from appendix: when does LSP egress delete bootstrapped BFD session?
[20:08:11] <Jared Mauch> question at mic: question use-case with monitoring ECMP with multiple sessions.
[20:09:34] <Jared Mauch> could be useful to track multiple paths...
[20:10:28] <Jared Mauch> BFD is continuity check
[20:11:40] <Jared Mauch> need to investigate multiple solutions w/ ECMP, LAG and other solutions.
[20:12:33] <Jared Mauch> at mic: Agree with previous speaker, not sure with what problem you are trying to solve
[20:13:12] <Jared Mauch> speaker: need for redundancy..
[20:13:49] <Jared Mauch> sam: (at mic) - how do you know where load balancing occurred, without adding entropy you can't measure this.
[20:15:38] <Jared Mauch> next steps: requesting comments from wg
[20:15:53] <Jared Mauch> jeff: lots of passion on this topic, bfd may be wrong hammer for this.
[20:16:20] <Jared Mauch> jeff: should IETF solve this problem (room consensus: yes)
[20:17:37] <Jared Mauch> Next Presentation: about fault monitoring for EVPN with BFD
[20:18:43] <adrianfarrel> BTW Jared, you're doing an excellent job. Thanks
[20:19:31] <Jared Mauch> (Trying)
[20:19:40] <Jared Mauch> (the cookies are good)
[20:20:06] <Jared Mauch> i believe this is the draft:
[20:20:07] <Jared Mauch> http://tools.ietf.org/html/draft-salam-l2vpn-evpn-oam-req-frmwk-02
[20:20:19] <Jared Mauch> er nope
[20:20:54] <Jared Mauch> Requesting comments from WG for fault monitoring of EVPN
[20:21:10] <Jared Mauch> Calling for adoption for WG
[20:21:33] <Jared Mauch> at mic: (Greg) - if multipoint BFD can be reused for this purpose
[20:22:36] <Jared Mauch> speaker: point-to-multipoint would apply, but not solve multipoint to point
[20:22:46] <Jared Mauch> this draft goes beyond p2mp draft
[20:23:24] <Jared Mauch> Jeff: Encouraging discussion on the list
[20:23:31] <Jared Mauch> and possible adoption in 1-2 months
[20:23:40] <Jared Mauch> Next:  Directed BFD return path
[20:24:15] <Jared Mauch> http://tools.ietf.org/html/draft-mirsky-mpls-bfd-directed-00
[20:28:02] <Jared Mauch> can use this to cause BFD to follow the same path both forward and reverse.
[20:28:41] <Jared Mauch> at mic( nobo- cisco) : measuring return path, great things to do
[20:29:12] <Jared Mauch> for segment routing, lsp ping, it will only capture one path from S-R (in diagram, source-dest)
[20:32:24] <Jared Mauch> at mic (myself:) discuss how NTT uses MPLS-TE and possibly LSP ping and how routing in diverse paths is perhaps desirable
[20:32:31] <Jared Mauch> at mic: Andrew Qu from Mediatech
[20:32:54] <Jared Mauch> BFD: More for continuity checking, for multihop operation is for more than link-level detection
[20:34:48] adrianfarrel leaves the room
[20:36:03] <Jared Mauch> Next presentation: BFD Proxy connections over monitored links - Brian Snyder
[20:36:53] <Jared Mauch> We work in satellite industry, just deployed layer-2 service over sat
[20:37:14] <Jared Mauch> keepalive checks alone very expensive for BGP endpoints
[20:37:27] <Jared Mauch> Looked at DLEP
[20:38:43] <Jared Mauch> DLEP doesn't scale, no vlan trunking, only prod up to ~70 nodes, vs 1k-10k nodes
[20:39:00] <Jared Mauch> BFD was better technology match than DLEP
[20:39:57] <Jared Mauch> by proxying connections it breaks auth, but links are already encrypted
[20:40:49] <Jared Mauch> use really high timers (we turned them off)
[20:43:10] <Jared Mauch> can inject packets with radios and not reply if link is down
[20:44:40] <Jared Mauch> plugged in and it worked in-place
[20:45:35] <Jared Mauch> would love time for longer timers.
[20:45:38] <Jared Mauch> no time for questions.
[20:45:58] <Jared Mauch> being presented tomorrow as well, or come talk to him..
[20:46:04] <Jared Mauch> (brian snyder)
[20:46:25] <Jared Mauch> Jeff: What is the goal, publish in BFD or other WG?
[20:47:24] <Jared Mauch> Next Presentation: use cases requiring new features of BFD
[20:47:37] <Jared Mauch> use case #1 - detection of multipath
[20:49:53] <Jared Mauch> use case #2 - application consistency
[20:50:58] <Jared Mauch> use case #3 - capability inquiry
[20:52:02] <Jared Mauch> use case #4 - state relay
[20:52:59] <Jared Mauch> use case #5 - detection of asymmetric LSPs
[20:53:21] <Jared Mauch> Questions:?
[20:53:41] <Jared Mauch> at mic: Greg Mirsky - keep discussion on list, including multipath
[20:53:59] <Jared Mauch> look at using IGP TE extensions to advertise BFD capability
[20:55:42] <Jared Mauch> Jeff: What is your goals with the draft
[20:56:07] <Jared Mauch> RFC?  Cause multiple solution drafts?  
[20:56:20] <Jared Mauch> Speaker: Goal for it to be informational draft for WG and derivative work from it
[20:56:35] <Jared Mauch> End of WG meeting
[20:56:45] noboakiya leaves the room
[20:57:27] Jared Mauch leaves the room
[20:57:30] Jared Mauch joins the room
[20:57:39] Jared Mauch leaves the room
[21:11:48] Jared Mauch joins the room
[21:14:53] Jared Mauch leaves the room
[21:14:54] Jared Mauch joins the room
[21:19:07] noboakiya joins the room
[21:23:40] noboakiya joins the room
[21:23:40] noboakiya leaves the room: Disconnected: closed
[21:24:04] noboakiya joins the room
[21:24:34] noboakiya joins the room
[21:25:23] noboakiya leaves the room
[21:31:59] malleus.errata leaves the room
[21:32:39] kiranmak leaves the room
[21:36:51] noboakiya leaves the room: Disconnected: closed
[21:40:33] noboakiya leaves the room: Disconnected: closed
[21:43:37] noboakiya joins the room
[22:04:28] adrianfarrel joins the room
[22:04:35] adrianfarrel leaves the room
[22:20:20] noboakiya leaves the room
[22:21:00] noboakiya joins the room
[22:21:52] noboakiya joins the room
[22:37:41] noboakiya leaves the room: Disconnected: closed
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!