Friday, July 25, 2014< ^ >
joel jaeggli has set the subject to: grow IETF89
Room Configuration
Room Occupants

[15:45:19] Rob Shakir joins the room
[15:45:30] Rob Shakir has set the subject to: GROW - IETF90
[15:52:56] internetplumber joins the room
[15:53:37] Wes George joins the room
[15:55:36] <Wes George> discussing
[15:55:43] <Wes George> agenda
[15:55:50] <Wes George> draft updates
[15:57:53] Samuel Weiler joins the room
[15:57:53] <Wes George> Jared now presenting
[15:58:50] <Wes George> slide routing resilience manifesto
[15:59:27] <Wes George> slide end product
[16:00:04] <Wes George> slide work in progress
[16:01:03] <Wes George> slide 7
[16:01:15] <Wes George> slide 8
[16:01:36] mikemlb joins the room
[16:01:43] <Wes George> slide 9
[16:03:10] <Wes George> slide 10
[16:04:14] <Wes George> slide 12
[16:04:50] Tony Tauber joins the room
[16:04:54] <Wes George> Jeff haas - resistance for routing registry
[16:05:04] <Wes George> home net potential for multi homed users
[16:05:29] <Wes George> jared - internet is global, if ISPs don’t care/don’t choose to take action, fine
[16:05:37] <Wes George> other regions seem not to have this issue
[16:05:50] <Wes George> need feedback on how to improve
[16:07:39] <Wes George> doug montgomery - concise and simply stated. what’s operative about it?
[16:08:01] <Wes George> what as an end user, do I understand about this
[16:08:42] mikemlb joins the room
[16:08:44] <Wes George> is there something here that is not present in the other docs like this?
[16:08:44] mikemlb leaves the room
[16:08:54] <Wes George> jared - should we publish a list of people who signed on?
[16:09:03] <Wes George> evolving doc that changes from year to year
[16:11:25] <Wes George> peter s - new RPSL info?
[16:11:44] <Wes George> jared - people try to treat this as proprietary, but this data is publicly available
[16:14:51] <Wes George> jared - some challenges are not fixable via RPKI and road
[16:16:14] <Wes George> andrei robachevsky
[16:16:17] <Wes George> john heasley
[16:17:18] <Wes George> ruediger volk
[16:20:24] <Wes George> doug montgomery
[16:20:46] <Wes George> if you say “register” provide guidance on where
[16:25:05] <Wes George> wes george, then andrei r at the mic
[16:26:30] <Wes George> Sriram now discussing
[16:28:23] <Wes George> slide 2
[16:29:41] <Wes George> slide 3
[16:34:05] <Wes George> slide 5
[16:34:38] <Wes George> slide 6
[16:36:05] <Wes George> slide 67
[16:36:07] <Wes George> 7
[16:38:34] <Wes George> slide 8
[16:38:51] <Wes George> slide 9
[16:39:12] <Wes George> slide 10
[16:40:04] <Wes George> slide 11
[16:42:27] <Wes George> slide 12
[16:43:16] <Wes George> slide 14
[16:44:25] <Wes George> slide 27
[16:44:28] <Wes George> er 17
[16:44:39] <Rob Shakir> slide++;
[16:45:50] <Wes George> slide 18
[16:47:18] <Wes George> slide 19
[16:48:12] <Wes George> slide 20
[16:49:03] <Wes George> slide 21
[16:49:53] <Wes George> slide 22
[16:53:22] <Wes George> Jeff haas
[16:53:36] <Wes George> worth leaving all solution spaces in here, remain in grow
[16:53:44] <Wes George> allows solutions to be evaluated along with problem statement
[16:54:48] <Wes George> nothing stops us from sticking this into optional transitive path attributes without BGPSec
[16:54:55] <Wes George> you could even put this in ext communities
[16:55:11] <Wes George> jakob heitz
[16:55:50] <Wes George> originator puts 2 attributes into the route. 1) you may pass this up and sign it, anyone can remove, cannot re-add
[16:56:02] <Wes George> 2) support the first attribute - backward compatibility
[16:56:30] <Wes George> doug M - we skipped over a bunch in attempt to be concise
[16:56:58] <Wes George> there’s a lot of discussion about what the properties need to be for a tag if you have a tag
[16:57:04] <Wes George> semantics
[16:57:33] <Wes George> protecting that aligns well with protecting BGP AS path
[16:57:51] <Wes George> community string, but you’d have to invent a new security method to protect communities
[16:58:27] <Wes George> joelja - agree’s with wes about separating drafts
[16:58:50] <Wes George> easier and less controversial to get a cogent definition of this
[16:59:03] <Wes George> we can explore solution space, they’re separable in terms of advance - different timelines
[16:59:38] <Wes George> yes, we can discuss both together, but not send to IESG that way
[17:00:05] <Wes George> andrei - how close is type 3 compared to valley-free routing?
[17:00:08] <Wes George> excluding peers?
[17:04:47] <Wes George> geoff huston - problem with route leaks - filters weren’t there, rpsl wasn’t filled out
[17:04:48] <Wes George> etc
[17:05:08] <Wes George> you’re suggesting a new thing that would help. if we all did the new thing, fixed
[17:05:24] <Wes George> a new tweak doesn’t fix “not everyone does this”
[17:06:33] <Wes George> probability of everyone doing RPKI is lower than the prob. of everyone doing route filters
[17:06:39] <Wes George> doug M
[17:07:16] <Wes George> Eric osterweil
[17:07:54] <Wes George> taxonomies are great, but need to reflect reality
[17:08:11] <Wes George> if you apply it to the corpus of data, do they work?
[17:08:17] <Wes George> are there things missing
[17:09:23] <Wes George> sandy - what do you measure to decide whether taxonomy is correct?
[17:09:50] <Wes George> eric - pick something that you’re going to demonstrate, show evidence
[17:10:53] <Wes George> sandy - “don’t send customer routes to non-customers”
[17:11:29] <Wes George> aside: can we stop saying “cogent” and “peering”? I’d like to keep my lunch down
[17:13:32] <Wes George> geoff - paper on detection of inter-as relationships spent a lot of time discussing how non-uniform they were
[17:13:48] <Wes George> relationships are fluid, time-dependent
[17:14:09] <Wes George> you don’t necessarily know, you’re guessing
[17:14:42] <Wes George> program @ georgia tech, grad student asking what happened when they saw an anomaly
[17:15:37] <Wes George> jared
[17:16:00] <Wes George> as part of the problem statement - vendors that are more prone to this type of thing than others
[17:16:10] <Wes George> do we document that?
[17:16:13] Samuel Weiler leaves the room
[17:16:23] <Wes George> protocol implementers, not ISPs-
[17:16:27] <Wes George> they default to send everything
[17:16:32] <Wes George> is that part of the problem?
[17:16:40] <Wes George> poor implementers/poor practices
[17:17:44] <Wes George> session ends
[17:17:48] Wes George leaves the room
[17:18:51] mikemlb leaves the room
[17:20:39] internetplumber leaves the room
[17:22:21] Rob Shakir leaves the room
[17:41:02] mikemlb joins the room
[17:41:31] mikemlb leaves the room
[18:01:06] Tony Tauber leaves the room
[18:01:06] Tony Tauber joins the room
[18:01:06] Tony Tauber leaves the room
[18:30:55] Tony Tauber joins the room
[19:08:34] Tony Tauber leaves the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!