IETF
idr
idr@jabber.ietf.org
Monday, 26 March 2012< ^ >
Room Configuration

GMT+0
[05:40:40] naikumar@cisco.com joins the room
[06:53:37] jlcJohn joins the room
[06:56:19] rjs joins the room
[06:58:55] Dave Freedman joins the room
[07:02:17] <Dave Freedman> anybody listening to the audio feed?
[07:04:24] <naikumar@cisco.com> yes...I can hear
[07:07:19] rjs leaves the room
[07:09:16] Stewart Bryant joins the room
[07:09:41] Wes George joins the room
[07:14:47] Florin Coras joins the room
[07:17:50] John Scudder joins the room
[07:25:07] jpc joins the room
[07:27:21] <Dave Freedman> any questions?
[07:28:54] rjs joins the room
[07:29:17] <John Scudder> can anyone hear the audio feed?
[07:29:39] Jeffrey Haas joins the room
[07:37:01] <John Scudder> I'll take that to mean either nobody is listening, or the audio feed is busted. hopefully the former.
[07:39:25] rjs leaves the room
[07:42:11] <jpc> audio works
[07:42:40] <Dave Freedman> great
[07:42:42] Florin Coras leaves the room
[07:46:55] rjs joins the room
[07:48:38] brian.peter.dickson joins the room
[07:50:02] brian.peter.dickson leaves the room
[07:57:33] Ilya Varlashkin joins the room
[07:57:51] Ilya Varlashkin is now known as iLya
[08:04:50] <Wes George> i'm bothered by the continual assumption that we *need* 50ms restoration
[08:05:00] <Wes George> that may have been true years ago, but anymore it's not realistic
[08:05:36] <Wes George> I'd like to see use cases that actually still require that short of a restoration time
[08:05:40] <rjs> The problem is, do you design for 50msec convergence, or do you design resilience in so that you have some means by which one doesn't rely on this restoration time?
[08:05:45] cbowers joins the room
[08:06:00] <Dave Freedman> well, it depends on whether you class service restoration as the *whole service*
[08:06:02] <Dave Freedman> or just a component
[08:06:27] <rjs> But the end user cares about the /whole service/ generally, they don't care *what* you had fail generally?
[08:06:28] <Wes George> you do both. there's talk in Opsarea about giving guidance to implementers about what's realistic for them to assume for convergence/restoration times
[08:06:48] <rjs> You do both to the extent that it is practical to do so, right?
[08:06:57] <Wes George> and then you try to get to the lowest convergence time before you get to that spike on the curve where your cost and complexity go near vertical
[08:07:01] <Dave Freedman> it really depends on what service you are designing for
[08:07:04] <rjs> And there are diminising returns in that practicality.
[08:07:08] <Dave Freedman> and what you can afford
[08:12:49] brian.peter.dickson joins the room
[08:20:45] gigix73 joins the room
[08:30:57] iLya leaves the room: I'm happy Miranda IM user. Get it at http://miranda-im.org/.
[08:31:17] Daniel King joins the room
[08:32:57] cbowers leaves the room
[08:33:50] ChrisBowers joins the room
[08:34:59] Stewart Bryant leaves the room
[08:41:00] <brian.peter.dickson> test message - please ignore
[08:41:14] <John Scudder> ignored!
[08:41:19] <Dave Freedman> heh
[08:41:31] <Dave Freedman> I think this is a good use case for operational , I'm going to ask him
[08:43:31] <brian.peter.dickson> I think remote just lost audio?
[08:43:47] <John Scudder> drat
[08:44:17] <John Scudder> i'll open a ticket and hope for the best
[08:44:28] <brian.peter.dickson> thx
[08:46:23] Stewart Bryant joins the room
[08:51:23] <Wes George> sorry, draft name is draft-gs-l3vpn-scaling
[08:51:49] <brian.peter.dickson> jgs - when Sandra is up, would you mind Skype-ing so I can hear? Thx
[08:52:38] <John Scudder> will do
[08:53:13] <John Scudder> I will bring up skype but turn down spkrs on my end
[08:54:52] weiler joins the room
[08:59:29] <John Scudder> audio feed being debugged now
[09:00:05] <brian.peter.dickson> may be moot for me, since I'm both next and last
[09:00:08] <Dave Freedman> I'm not sure this should be an IDR problem
[09:00:20] <Dave Freedman> this should be a GROW problem
[09:00:21] <Dave Freedman> no?
[09:01:28] <John Scudder> brian, do you want me to mute my mic when you start talking?
[09:01:37] <John Scudder> or leave it open?
[09:02:16] <brian.peter.dickson> yes, please - until questions at end? am using laptop mic/speakers built-in
[09:02:23] <John Scudder> okdk
[09:02:44] <brian.peter.dickson> Can you chat any audio related feedback - too quiet, too loud etc?
[09:03:35] <John Scudder> will try
[09:08:13] <Wes George> Dave - I am not opposed to seeing it in GROW also, but I think IDR has to be involved for protocol changes. defining what a leak is might be grow, then IDR for how to signal it, SIDR to protect it
[09:08:59] <Dave Freedman> wes: exactly
[09:09:12] <Dave Freedman> I want GROW to make the spec for what we believe a leak to be
[09:10:07] <Wes George> I'd recommend cross-posting that to the three lists
[09:10:53] <Jeffrey Haas> What RPSL doesn't do well is dealing with extremely large transitive graphs
[09:10:58] gigix73 leaves the room
[09:11:08] Dave Freedman leaves the room
[09:11:42] <Wes George> well, that and there are folks who don't participate/don't want to participate in publishing all of fthat info
[09:12:04] <Jeffrey Haas> If you're going to put this in SIDR, *and be enforced*, it has to be published.
[09:12:25] <Jeffrey Haas> I'm not sure there's a lot of BGP work to be done. The coloring is a "community" style problem. RPSL is *mostly* good enough for describing the policy.
[09:12:39] <Wes George> yes, for some value/method of publish
[09:12:45] <Jeffrey Haas> Jabber is publishing my messages out of order. :-(
[09:12:49] Dave Freedman joins the room
[09:13:04] <Wes George> I was specifically referring to publishingg in some sort of public database like RADB
[09:13:05] Dave Freedman leaves the room
[09:13:15] Padma Pillay-Esnault joins the room
[09:13:43] <Jeffrey Haas> RPSL is mostly sufficient. But describing the transitive AS relationships, much less link-specific roles, is not really scalable.
[09:13:59] gigix73 joins the room
[09:14:14] gigix73 leaves the room
[09:14:26] <weiler> should we name the customer Mikey?
[09:14:49] gigix73 joins the room
[09:17:49] Dave Freedman joins the room
[09:19:38] rjs leaves the room
[09:19:43] rjs joins the room
[09:25:02] Daniel King leaves the room
[09:27:07] rjs leaves the room
[09:27:26] <John Scudder> 3 minute warning
[09:28:26] <John Scudder> what slide now?
[09:30:49] <John Scudder> Time is up so please wrap as soon as you feasibly can.
[09:33:22] weiler leaves the room
[09:33:39] Jeffrey Haas leaves the room
[09:33:57] Wes George leaves the room
[09:34:04] <John Scudder> can you hear q's ok?
[09:34:10] <brian.peter.dickson> yes
[09:37:20] <brian.peter.dickson> just lost sound
[09:38:18] weiler joins the room
[09:38:36] Stewart Bryant leaves the room
[09:39:06] weiler leaves the room
[09:41:37] gigix73 leaves the room
[09:42:20] jpc leaves the room
[09:44:35] brian.peter.dickson leaves the room
[09:45:30] Dave Freedman leaves the room
[09:46:58] ChrisBowers leaves the room
[09:51:28] John Scudder leaves the room
[10:14:00] Padma Pillay-Esnault leaves the room
[10:50:00] jlcJohn leaves the room
[10:50:46] Dave Freedman joins the room
[11:01:39] ChrisBowers joins the room
[11:01:41] ChrisBowers leaves the room
[11:04:21] Stewart Bryant joins the room
[11:11:14] Dave Freedman leaves the room
[11:17:07] Jeffrey Haas joins the room
[11:17:15] Jeffrey Haas leaves the room
[11:18:14] Dave Freedman joins the room
[11:31:31] Dave Freedman leaves the room
[13:05:15] Stewart Bryant leaves the room
[13:20:45] Stewart Bryant joins the room
[14:05:39] Stewart Bryant leaves the room
[15:05:52] Stewart Bryant joins the room
[15:15:09] naikumar@cisco.com leaves the room
[15:41:05] Stewart Bryant leaves the room
[15:56:16] Stewart Bryant joins the room
[16:47:25] Stewart Bryant leaves the room
[19:37:18] Stewart Bryant joins the room
[20:28:20] Stewart Bryant leaves the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!