[02:58:17] behcet.sarikaya joins the room [04:01:20] marshall joins the room [04:01:30] I am going to be the jabber scribe [04:01:49] The FIRST meeting of multimob [04:02:30] agenda bashing [04:02:38] priority is the charter item discussions [04:02:52] existing milestones [04:02:58] two items [04:03:14] a document explaining multicast in pmip-6 [04:03:29] and a document on how to tune igmpv3/mldv2 [04:04:00] next is Stig on the charter [04:05:29] Stig : the charter right now is fairly limited [04:05:35] basic multicast support [04:05:44] shamus joins the room [04:05:45] once we do that, we hope to work on extensions [04:06:04] once the basic solution is done we can extend the charter [04:06:38] Jari : You do the basic stuff, do it fast, do it well, you get to work on the extensions [04:06:57] Stig : What do we mean when we talk about the basics [04:07:01] ? [04:07:10] protocol extensions are out of scopy [04:07:12] scope [04:07:45] need to show how to do multicast without any changes to PMIP6 [04:08:02] also, should not require support or changes in the local network [04:08:21] Milestones [04:08:38] how to do pmip6 multicast is due now [04:09:15] Hitoshi is next [04:11:04] this idea is how to optimize igmpv3 for multicast mobility [04:11:37] we need to minimize power consumption and the number of messages sent [04:11:58] I, personnaly, cannot understand optimization without protocol modification [04:12:29] wireless is really sensitive to the situation [04:12:47] static values are too optimistic [04:12:57] proposals [04:13:07] tracking of membership status [04:13:43] DCN-0D4B06D5DF0 joins the room [04:13:48] explicit tracking function as defined in 3376 and 3810 [04:14:17] router can recognize what kind of receivers are there [04:15:05] the cons are that router needs processing power and large memory - [04:15:10] no timer change [04:15:22] General Query Processing [04:16:22] general query is unicasted to recorded members and multicasted\ to refresh memory [04:16:34] con : longer leave latency [04:16:42] long join latency [04:16:58] this requires new timer values [04:17:20] cannot mix unicast and multicast for general query [04:17:32] we may need to move this to the extension draft [04:17:43] IGMP report processing [04:18:23] igmp allows for 0.0.0.0 (unspecified) addrress [04:18:31] mld does not [04:18:52] this is only for explicit tracking\ [04:19:00] - multicast source filter [04:19:32] this is the recomendation of the IGMPv 3 Lightweight people [04:21:09] he does not think that there is any reason not to do this [04:21:44] Last proposal : Excplicit Membership Notification - will have to be moved to the extension draft [04:22:21] DCN-0D4B06D5DF0 leaves the room [04:22:31] new timer value and new behavior is required [04:23:36] so, the revised plan does not have the Query processing and explicit membership notification [04:25:22] Questions ? [04:25:25] Stig [04:25:45] Stig : The draft is interesting [04:25:57] The charter is strict [04:26:19] we need to split this into a charter draft and a non charter draft [04:27:09] about the timers - I agree that it will be hard to get one set of static times [04:27:12] timers [04:28:13] Hitoshi : that is true, but it is really beneficial [04:28:38] Thomas : THere is also the retransmit value [04:29:01] there is no perfect value [04:29:33] shamus leaves the room [04:29:49] Suresh : Another comment - it will be very useful to say how changing the value for the timer changes the behavior [04:31:02] Hitoshi : Mobile node usual do not move (?) [04:32:43] Stig : MLD says that multicast must support unicast general queries [04:33:07] but I don't know of any routers that support this [04:33:25] it might be ok to say that a router could chose to do that [04:34:39] Thomas : MLD-lite - the model node might want to use an exclude message to exclude a code [04:35:57] Stig : I don't know of any application that makes use of this. It's not clear to me if we should recommend that. [04:37:42] Next is evaluation of the charter [04:38:00] recommend a set of timer values [04:38:08] there is Hitoshi draft [04:39:00] Stig : Optimizations discussed - unicast general queries [04:40:05] Marshall : Is this still in the lightwieght version ? [04:40:12] Stig : I believe it does [04:41:02] Hitoshi : the draft should be changed from optimization to requirement [04:43:26] Suresh - it's not just the nodes but also the router - I don't think that unicast general queries are supported either [04:43:56] Stig the other open question is the lightweight stuff [04:44:20] it's great to simplify the model nodes [04:44:48] Jari : There are multiple things that the charter could mean [04:45:30] for example, we could profile the mobile environment, and that could include lightweight [04:45:36] cjbernardos joins the room [04:46:13] we could have a very short document saying "we can improve things without changing protocols" and that's ok [04:50:56] Marshall : We have debated the lightweight IGMP a long time in MBONED [04:51:55] while it doesn't change the protocol on the wire, it actually does, because it means that the exclude mode can never be used [04:52:19] what it really does is to recongize that the protocol in fact has already changed [04:53:13] Stig : can we adopt Hitoshi's document as the basis on theIGMP wg item ? [04:53:26] [apparent consensus] [04:53:37] ok, let's move forward on that basis. [04:55:01] Thomas : the idea of this draft is to bring multicast into PMIP world as we know it [04:55:09] use lightwieght standard compoinents [04:55:27] we want as much aggregation as possible [04:56:00] in PMIP there is a 1:1 association between a LMA and a mobile node [04:56:21] how do put multicast in this [04:56:33] ? [04:56:43] put a standard multicast querier on the LMA [04:56:57] put a standard MLD proxy on the MAG [04:58:40] outcome : We believe that this is a very natural proposal [04:58:47] mld proxies are well adopted [04:59:00] traffic flows may not be optimal [04:59:36] two mobile nodes with different LMAs but the same MAG will mean duplicate traffic [04:59:53] native multicast to the mags will help with this [05:00:46] NEXT is Suresh [05:01:49] Suresh : this draft was written before the charter came out - this does require small changes in behavior AT THE MAG [05:02:04] idea - make the LMA a multicast router [05:02:20] a packet could have 3 headers [05:02:49] all the multicast packets are tunneled to the MAG [05:03:23] Marshall : Is there any aggregation in this solution ? [05:03:26] Suresh : No [05:03:40] Next is Seil Jeon : [05:04:02] two proposals [05:04:15] multicast router is collocated with an LMA [05:04:24] or its totally separate [05:05:23] in their solution only the MAG has a MLD forwarding proxy function [05:06:02] need a fast handover [05:06:51] we think that enhanced skill is required to do this [05:08:11] Stig : Can this proposal be merged with Thomas's ? [05:08:16] Seil : Yes [05:08:25] We have already discussed this [05:09:17] Next is Guang Lu [05:09:42] Guang Lu : We use the existing procudures as mch as possible [05:09:53] the MAG is a multicast process [05:10:17] dedicated multicast LMA [05:10:20] LMA-M [05:10:30] MAG connects to both LMAs [05:11:00] zunigajc joins the room [05:11:07] zunigajc leaves the room [05:12:07] this removes the tunnel convergance issue at the MAG [05:12:10] zunigajc joins the room [05:17:54] Marshall : Asked about whether you need multiple M-LMA's [05:17:58] apparently not [05:20:26] Carlos Bernados : [05:20:53] Key pints - MAG maintains the MN multicast status and is the multicast first hop [05:21:03] the MAG can be a MLD proxy [05:21:16] or a multicast router [05:27:07] Thomas : If a mobile node is multihomed and wants to have mobility on all interfaces, this is beyond what we are doing [05:28:42] Behcet : Multi interface multicast is out of the charter [05:29:06] Bernados : PMIP already supports multiple interfaces [05:32:08] Stig : the question is whether the charter allows options that are modifications [05:32:33] ? : Do solutions cover uplink multicast ? [05:32:40] Behcet ; it's out of scope [05:32:46] multicast receiver only [05:34:39] Stig : We have two basic solutions in the MAG [05:34:51] MLD proxy or multicast at the MAG [05:35:01] or do the tunneling from the LMA [05:35:17] either you keep state in the MAG or the MLD [05:35:30] I like the MLD proxy solution [05:35:49] I am afraid of the duplication on the Christian proposal [05:38:56] We don't have one solution fits all ? [05:43:19] ipv4 support [05:45:13] Thomas : I don't see what proxy solution is different between v4 and v6 [05:45:22] Stig : I don't see this as a problem [05:45:38] [Marshall : I don't either] [05:54:21] discussion about whether or not a new interface triggers a query [05:54:44] I don't think that the protocol requires this at all [06:00:32] Show of hands [06:00:52] How many people believe that we should do a LMA solution [06:01:08] [06:01:28] How many people believe we should pursue the IMGP proxy at the MAG [06:01:31] <16> [06:01:42] How many people believe it is too soon to decide [06:01:44] <2> [06:03:32] zunigajc leaves the room [06:04:46] cjbernardos leaves the room [06:04:58] behcet.sarikaya leaves the room [06:05:25] marshall leaves the room