[06:58:11] --- SharonChisholm has joined
[07:27:12] --- SharonChisholm has left
[07:34:27] --- bruce has joined
[07:51:37] --- javier has joined
[07:55:38] --- SharonChisholm has joined
[07:58:28] <SharonChisholm> Is someone scribing?
[08:00:15] <bruce> nyet; not physically there.
[08:00:35] <SharonChisholm> <ok ... I'll do a bit>
[08:00:42] <SharonChisholm> We are going through document status
[08:01:07] <SharonChisholm> ------------
[08:01:20] <SharonChisholm> Advertising equal cost multi-path (ECMP) routes in BGP
[08:01:36] <SharonChisholm> - last presented in vienna
[08:01:43] <SharonChisholm> - diffs from last time
[08:02:12] --- torus has joined
[08:02:28] <SharonChisholm> intro - providers install ECMP routes, announce just one. This can break routing, policies, etc.
[08:02:49] <SharonChisholm> if 2 routes, advertise one, routing different than forwarding
[08:03:00] <SharonChisholm> <picture>
[08:03:32] <SharonChisholm> <the LCD projector is too close to the screen ... not getting whole picture>
[08:03:49] <SharonChisholm> <or is that too far away?>
[08:04:12] --- avri has joined
[08:05:16] <bruce> <too close if its off the boundaries of the screen, too far away if theres sizeable black space around the border, and too high a resolution if the presenter keeps making remarks about 'it doesn't show up, but on my screen its really good'>
[08:05:42] <SharonChisholm> <too close then>
[08:06:19] <SharonChisholm> <there is detailed traffic being routed in his walk through the picture>
[08:06:45] <SharonChisholm> What if R2 and R3 don't support ECM extensions?
[08:06:58] <SharonChisholm> <alex fixed projector :-) >
[08:07:14] <SharonChisholm> Contructing synthetic AS_PATHs ...
[08:07:23] <SharonChisholm> splitting load accross the two AS PATHs
[08:07:52] <SharonChisholm> AS_PATH1_PATH1:X Z { A B C }
[08:08:13] <SharonChisholm> AS_PATH2: X1 Z1 XZ1
[08:08:25] <SharonChisholm> AS_PATH1: X Z { A B C }
[08:09:06] <SharonChisholm> AS_PATH3 : X2 Z2 Z3
[08:09:16] <SharonChisholm> various ways to build out an AS PATH
[08:09:21] <SharonChisholm> With BGP ECMP ...
[08:09:32] <SharonChisholm> - can avoiud sub-optimal routing with route reflectors
[08:09:41] <SharonChisholm> - avoice some cases of persistent route ocsillations
[08:09:53] <SharonChisholm> - load balance BGP!
[08:10:01] <SharonChisholm> How to deal with ...
[08:10:06] <SharonChisholm> - multiprotocol extensions
[08:10:17] <SharonChisholm> - BGP route flap dampening
[08:10:24] <SharonChisholm> - BGP/MPLS VPNSs
[08:10:30] <SharonChisholm> - regular expressions
[08:10:41] <SharonChisholm> - Split load accross ASes
[08:11:20] <SharonChisholm> current implimentations look into AS sequences ...
[08:11:35] <SharonChisholm> some people are concerned about OAM part
[08:11:53] <SharonChisholm> don't want to split 'case it will make OAM difficult
[08:12:15] <SharonChisholm> most routers are flow based .... it means you won't get your packets going on different streams
[08:12:21] <SharonChisholm> Next Steps
[08:12:24] <SharonChisholm> - operator feedback
[08:12:28] <SharonChisholm> - making it a wg doc
[08:12:36] <SharonChisholm> Chair - how many read the document
[08:12:51] <SharonChisholm> chair - out of those who read it ... how many think it should be a wg document
[08:13:11] <SharonChisholm> <one person from what I could see but I don't think the number read was huge either >
[08:13:26] <SharonChisholm> chair - move to working group mailing list
[08:13:51] <SharonChisholm> ------------------------------------
[08:14:02] <SharonChisholm> draft-ck-bgp-atm-nlrt-00.txt
[08:14:19] <SharonChisholm> Not going to discuss draft, but rather a liaison from some MPLS group
[08:14:33] <SharonChisholm> MPLS-FR Alliance
[08:14:51] --- javier has left: Disconnected
[08:15:00] <SharonChisholm> <flow chart of interworking between IDR WG, L2VPN WG and MPLS-FR Alliance>
[08:15:28] <SharonChisholm> liaison was sent to the mailing list a couple weeks ago
[08:15:45] <SharonChisholm> this draft is good ... please progress to RFC
[08:16:22] <SharonChisholm> chair - couple questions .. first to ADs ... let's assume first that the wg accepts ... is it ok from the IESG to access something as result MPLS-FR request?
[08:16:28] <SharonChisholm> alex - don't understand the question
[08:16:39] <SharonChisholm> chair - <refers to flow chart>
[08:17:10] <SharonChisholm> alex - just input from the frame relay aliance is not enough ... ther needs to be enough wg interest plus the rechartering bit ...
[08:17:21] <SharonChisholm> chair - so, first poll the wg and then talk about recharter
[08:17:31] <SharonChisholm> chair - so, we will take to mailing list ...
[08:17:45] <SharonChisholm> chair - how many read ... 20 ish .. how many think it should be a working group document ... 5ish
[08:17:59] <SharonChisholm> Q - if we do this, will the work be done in both bodies?
[08:18:18] <SharonChisholm> A - yes, we have been working on this for a while over there ... our document is more extensive ...
[08:18:25] <SharonChisholm> Q - so the overlap is not extensive?
[08:18:28] <SharonChisholm> A - no ...
[08:19:03] <SharonChisholm> chair - peter why don't you send an email to the idr working group asking them to accept .,.. if we get concensus from working group we will pass to IESG for rechartering
[08:19:07] <SharonChisholm> --------------------------------------
[08:20:14] <SharonChisholm> Connecting IPv6 islands over ipv4 MPLS using IPv6 provider endge routers (6RE)
[08:20:19] <SharonChisholm> Francois
[08:20:32] <SharonChisholm> Problem statement
[08:20:46] --- yushunwa has joined
[08:20:57] <SharonChisholm> - operator offers IPv4 services (IPv4 internet, ipv4 vpns ) ...
[08:20:59] <SharonChisholm> - something else
[08:21:12] <SharonChisholm> - operator runs an ipv4 mpls core
[08:21:18] <SharonChisholm> - operator wants to offer ....
[08:21:30] <SharonChisholm> solution is what is defined in ID
[08:21:45] <SharonChisholm> how to combine existing things to acheive what we need to do
[08:22:01] <SharonChisholm> same architecture as 2547 from high level perspective
[08:23:11] <SharonChisholm> embedd ipv4 address using existing method of embedding that into ipv6
[08:23:17] <SharonChisholm> background
[08:23:27] <SharonChisholm> - long history ... this is the third working group
[08:25:41] <SharonChisholm> Changes since -03
[08:25:50] <SharonChisholm> - cleaned up usage of must/should/may
[08:26:00] <SharonChisholm> - expanded inter-as scenarios
[08:26:08] <SharonChisholm> expanded security considerations
[08:27:03] <SharonChisholm> Relationship with l3vpn-bgp-ipv6
[08:27:18] <SharonChisholm> - tunnel document specifies global ipv6 reachability
[08:27:40] <SharonChisholm> - l3vpn bgp specificies BGP/MPLS IPv6 VPNS
[08:27:49] <SharonChisholm> - both documents are consistent where they can be
[08:27:56] <SharonChisholm> - docuiments differ were needed
[08:28:13] <SharonChisholm> -- example local SAFI versus VPN SAFI, etc.
[08:28:27] <SharonChisholm> Next steps - issue last call ... conduct implementation survey
[08:29:29] <SharonChisholm> ------------------------------------------------
[08:29:30] --- javier has joined
[08:30:02] <SharonChisholm> AS Confederations Edge (Sue)
[08:30:11] <SharonChisholm> There is no draft since we thought of this after the cut off ... stuff sent to mailing list
[08:30:19] <SharonChisholm> this is a particular scenario in the network
[08:30:27] <SharonChisholm> as confederation topology
[08:30:35] <SharonChisholm> - this works on paper at the moment
[08:31:29] <SharonChisholm> Topology ... linked in a ring ... might happen in satelite
[08:31:41] <SharonChisholm> stuff on the ground ... not ring
[08:32:16] <SharonChisholm> what happends is the links drop out and the guy (between the two topologies) is alone ... ideally it would be good if things kept going
[08:32:30] <SharonChisholm> Would like to fall back ... reset BGP connection to another AS
[08:32:37] <SharonChisholm> What's new?
[08:32:48] <SharonChisholm> - if do via configuration, you drop the port
[08:33:08] <SharonChisholm> - make withiout dropping for AS confederation nodes using dynamic capability
[08:33:24] <SharonChisholm> --- open echange sets it up
[08:33:39] <SharonChisholm> open cabablies ...
[08:34:54] <SharonChisholm> send prefix flags - resend routes in groups; send all routes immediately; don't resend any routes
[08:35:40] <SharonChisholm> Dynamic capabilites
[08:35:55] <SharonChisholm> - index into list for AS in use ....
[08:36:23] <SharonChisholm> example from A mobile AS>
[08:37:10] <SharonChisholm> same topology but it recovers from the link failure
[08:37:34] <SharonChisholm> people arn't yet seeing problems ... this is simple ... that's not always good
[08:38:32] <SharonChisholm> Blue (R) sheets?
[08:39:21] <SharonChisholm> Mike - if you expect topology to change ... why use confederations? Why not usual BGP so you don't have to change your personality?
[08:39:36] <SharonChisholm> Peter - why not an IGP?
[08:39:43] <SharonChisholm> Sue - polices issues
[08:40:11] <SharonChisholm> Mike - obersevation ... you can't nest confederations from a technical perspective
[08:40:20] <SharonChisholm> sue - nested or interlocking
[08:40:30] <SharonChisholm> mike - neither .. i can show you more offline
[08:40:49] <SharonChisholm> mike - third thing ... for the working group ... do we want to turn BGP into a mobile adhoc networking thing.
[08:40:58] <SharonChisholm> sue - not defined as mobiled i <somthing>
[08:41:12] <SharonChisholm> sue - not trying to go there ... the term is mobile AS ... that is the IRTF discussion
[08:41:31] <SharonChisholm> sue - mobile adhoc, mobile as and mobile IP ... different things
[08:41:48] <SharonChisholm> sue - i don't share the same definition of adhoc
[08:42:02] <SharonChisholm> mike - if we do want to go here .. need to look in broader context
[08:42:29] <SharonChisholm> cisco - two comments. in your talk ... wrong level of granularity ... <stuff> ... too small .. unless you are talking about large groups
[08:42:39] <SharonChisholm> sue - example is small so it fits on the screen
[08:42:55] <SharonChisholm> cisco - not sure what the added security buys you ....
[08:43:39] <SharonChisholm> sue - what I was aiming for ... see if I can ... first is TCP does not encrypt ... authen the other side. not dangerous thigns ... could be DOS attack ... do we need something on top of that ... that is all that is ...
[08:44:07] <SharonChisholm> Q - Is the policy for the links or to the AS
[08:44:13] <SharonChisholm> sue - to the conferenation or somthing else?
[08:44:22] <SharonChisholm> Q - no, is the policy for the link or the group of routers?
[08:44:24] <SharonChisholm> sue - which one
[08:44:40] <SharonChisholm> sue - 501 , 500 and 502 ....
[08:44:51] <SharonChisholm> Sue - policies used to reduce iBGP mesh
[08:45:10] <SharonChisholm> Q (peter) - saving some stuff,. but you have to reroute all the data
[08:46:05] <SharonChisholm> Q2 - the security bit, not so sure of ... perhaps we should leave it at transport layer. the reference to adhoc .. i would ask the people thinking of this to remove and consider it a general case and see if it makes sense to solve it in the general case
[08:46:33] <SharonChisholm> john - i think it is a good characterization ... a confed that might get broken into two pieces
[08:46:47] <SharonChisholm> chair - actuallly, the current spec talks about partitioned ASes
[08:46:52] <SharonChisholm> john - I know ;-)
[08:47:05] <SharonChisholm> sue - any more feedback?
[08:47:36] <SharonChisholm> MCI - you put boxes in confed, usually it is the same igp ... <something about reset> ...a re we assuming all the boxes are in iBGP
[08:47:54] <SharonChisholm> sue - so not all BGP environments have IPBGs
[08:48:15] <SharonChisholm> mCI - different ... whether next hop gets reset ...
[08:50:14] <SharonChisholm> peter - are we trying to see what we can do with BGP or to solve the problem?
[08:50:18] <SharonChisholm> sue - to solve the problem
[08:50:36] <SharonChisholm> -------------
[08:50:50] <SharonChisholm> chair - issue from mailing list on existing RFC .... people get involved
[08:51:08] <SharonChisholm> chair - we are done at this point ...
[08:51:19] <SharonChisholm> (S)
[08:52:47] --- javier has left
[08:53:15] --- torus has left
[09:09:24] --- SharonChisholm has left
[09:43:53] --- bruce has left
[10:27:08] --- avri has left: Disconnected
[10:33:28] --- avri has joined
[10:50:30] --- yushunwa has left: Disconnected
[11:54:54] --- avri has left: Disconnected
[12:09:55] --- avri has joined
[13:56:54] --- avri has left: Disconnected
[14:26:25] --- avri has joined
[15:44:47] --- avri has left: Disconnected