[00:28:23] Stewart Bryant leaves the room [00:55:24] Stewart Bryant joins the room [01:27:55] Stewart Bryant leaves the room [06:18:00] Stewart Bryant joins the room [08:21:37] Stewart Bryant leaves the room [14:51:07] Stewart Bryant joins the room [15:48:58] Stewart Bryant leaves the room [16:02:52] YJS joins the room [16:02:56] David Sinicrope joins the room [16:03:17] Hi everybody - took some time to get on jabber [16:03:58] David Black talking - with Linda Dunbar working on cleaning up the fiber channel draft [16:04:05] Linda did a lot of work [16:04:23] Stewart Bryant joins the room [16:04:35] original authors were "insufficiently incompetent" [16:04:47] other SDOs are dependent on this draft, so need to finish [16:05:15] we are going to removed FC-flow draft, so back to straightforward PW [16:05:59] FC PW will take advantage of reliable transport provided by MPLS-TP instead of TCP-like [16:06:05] however strong delay constraints [16:06:30] differentiate FC PW more from FC IP [16:06:41] David will have to work back in the FC SDO [16:06:59] many many changes - won't go through all of it (housekeeping) [16:07:27] getting the draft aligned with how FC is now used\ [16:07:51] will be ready for another WG LC before next IETF [16:08:18] Shahram - reliability in MPLS-TP ? - who says TP is reliable ? [16:08:58] David - TCP friendliness is overkill, here no retransmissions by PW, rather FC can cope with occassional loss [16:09:10] use ASFC (like GFP-T) [16:09:26] Stewart - we reserved 3 codepoints - are these still needed ? [16:10:15] David - previous slide says reserved code points for SR, should leave just in case someone is using [16:10:41] Stewart- make IANA note that reserved for history [16:11:18] David - the codepoints used for SR should be marked not for reuse [16:11:54] next - Luca - no its George - no-one is coming up [16:12:02] Matthew is presenting [16:12:13] Pseudowire Status For Static Pseudowires and MPLS-TP enhancement draft-ietf-pwe3-static-pw-status-00.txt [16:12:47] sorry for confusion about presenter - this is about protocol design for status when no control plane [16:13:02] Replace LDP transport layer with a data-plane based transport layer [16:13:11] Simulate session by periodic message transmission [16:13:27] acknowledge option added to quench the PW status transmitter [16:13:35] also is S-PE Bypass option for AC status [16:13:46] showing bits for message [16:14:16] what has been added to draft is a mechanism for static PW status across ACH [16:14:39] default value for sending is every 30 seconds [16:14:51] optional ACK from remote PE [16:15:03] stop sending when fault cleared [16:15:08] scaling issue [16:15:26] requirements - need to validate status for ALL PWs [16:15:41] need to protet againt lockup and reduce message rate [16:15:51] showing status verification [16:15:57] please read draft and comment to list\ [16:16:13] asking if need to add refresh reduction option\ [16:16:58] question - can use if there IS an LDP session ? [16:17:10] still useful [16:17:22] Matthew - intention was NOT to use [16:22:00] adrianfarrel joins the room [16:22:03] Fei Zhang - LDP Extensions for MPLS-TP PW OAM configuration [16:22:25] redefines PW OAM config TLV [16:23:35] defines new flags for OAM - alarms enabled, desired MEP/MIP entities [16:24:17] showing how to set up OAM using sub TLV in signaling [16:24:32] now showing how to delete OAM entities [16:24:59] need to delete both source and sink entities [16:25:37] next steps - need to handle packet loss/delay configurations [16:25:42] keep track of TCM [16:36:45] lots of talk at mike - sorry I was there so couldn't jabber [16:37:26] Lou Berger - MPLS-TP Control Plane Framework (just mentioned at mike) [16:37:52] for those who haven't been following - mapping requirements for control plane from all the TP documents [16:38:11] enumerate reqs from all docs that impact control plane [16:38:28] mostly cut and paste but sometimes MUST becomes SHOULD or MAY [16:38:51] this provides context for the rest of the document - how to meet using PW and MPLS-TP mechanisms\ [16:39:35] TP does not require control plane, PWE has one - this drives which functions are in OAM and which are where [16:40:15] weihongbo joins the room [16:40:49] after enumeration gaps were found, and these have to handled either in MPLS or here [16:41:01] PWE people - please read draft [16:41:28] this ID was submitted as 01 WG draft in CCAMP [16:41:37] Eric fixed up for 01 [16:42:20] note that this is a CCAMP doc, but will be presented to both [16:43:11] after comments will respin in April [16:46:18] curtis at mike - don't want to split this into two documents - one in CCAMP and one in PWE [16:46:38] Malcolm - appropriate to have one control plane document for both LSPs and PWs [16:51:06] Hello, which slide now? [16:52:04] now slide - all talk at mike [16:52:20] speaking about separate documents [16:52:31] got it, thanks [16:53:36] next - leftover drafts from MPLS session yesterday [16:54:40] George is going to speak about TP identifiers draft [16:54:50] Italo sent comments and they are being incorporated [16:54:57] very important that people review before LC [16:55:17] there is a half-open issue of reconciliation with OAM framework [16:55:45] I think it's Very Important draft [16:56:57] George continues - loopback draft [16:57:05] there was a meeting yesterday [16:57:24] no slide? [16:57:29] now only about lock construct and heartbeat [16:57:29] No webex? [16:57:31] no slides [16:57:37] ok [16:58:26] Tae-sik Cheung on TP mesg protection (there ARE slides) [16:59:15] survivability is important, TP requires efficiency, so mesh protection enabling sharing of resources [16:59:35] 1:n can meet requirement, since more efficient than 1:1 [17:00:42] require interdomain coordination mechanism [17:01:03] showing how requirements from 5654 map to this solution [17:01:40] summary - builds in linear protection draft, but extends for resource sharing [17:01:51] the draft has procedures and protocol [17:02:12] next steps - need manageability and security [17:02:19] please review and comment [17:03:03] next Yuanlin Bao - TP path transfer [17:03:51] some SPs have already deployed TP for mobile backhaul, but management plane configuration [17:04:32] may want in future control plane, and ownership transfer should not impact traffic [17:05:01] req from 5654, CCAMP addressed this for LSP [17:05:08] what about PW ? [17:05:36] weihongbo leaves the room [17:06:18] quit LDP PW session, and then no label mapping, so need to extend LDP [17:07:44] asking for this to become WG doc [17:18:53] still discussion at mike - most people saying that this is not needed [17:19:20] people who want it are saying that people are not saying that it is not needed, rather that is is not possible [17:19:32] maintenance windows no longer exist [17:19:45] George saying that hasn't heard business justification [17:20:20] this would be a major re-engineering of how MPLS works [17:20:54] Stewart - no one says that need protocol action - this is a local matter [17:22:19] Loa - there is make before break already defined [17:22:50] since TP work is already congested, let's not take on new work [17:24:00] George - can maintain traffic without keeping the same label [17:24:30] Lou - there is already a doc for LSPs, this is no req for this for PWs [17:25:39] next, same presenter continues about framework for TP ownership [17:25:51] req in 5654 [17:26:16] this doc extends GMPLS control plane for ownership transfer [17:26:45] need to extend LDP - set up and delete between T-PEs [17:27:14] showing how works in SS and MS PWs [17:27:57] explaining how do for association for bidirectional LSPs [17:28:02] showing this scenario [17:28:39] next steps - please submit comments so can refine [17:34:29] next up - Sami - Time-to-Live TLV for LSP-Ping [17:34:50] extension to LSP ping (NOT TP draft !) [17:35:03] motivation is to verify connectivity of a segment of a Multi-Segment PseudoWire (MS-PW) from any node on the path of the MS-PW. [17:35:37] problem definition - MS-PW spans segments, and want to ping from stitching point to stitching point [17:35:49] need TLV to tell the receiver the number of hop delays [17:36:08] so add to echo req a TTL TLV [17:37:21] show format of TLV to be added (after timestamps ?) [17:38:05] Greg - applicable to associated channel ping too [17:38:52] Dave A - greg's proposal makes MIPs and MEPs the same [17:40:29] George - still MEP to MIP, but for MS-PW need to ping back from last LSR in transit domain [17:41:07] Dave- but then TTL of sender needs to be the same as outgoing [17:41:32] George - S-PEs are always co-routed [17:44:04] last talk - Stewart about TLV draft in LC, and would like to gauge what WG wants to do with some of the comments received [17:50:20] experiencing technical difficulties - please standby [17:51:00] great that 5 telecommunications experts are working on this [17:51:44] finally we are up and running [17:52:27] TP ACH TLV draft - let's take care of this now, so we can resubmit ASAP [17:52:55] 1st comment - add experiemental TLVs (will add 8 values) [17:53:06] probably top 8 [17:53:32] dest address - some asked for this, easy to add (2 new TLV types) [17:54:01] but need IPv4 and IPv6 ? none of the OAM protocols need them [17:54:35] after this ID is published new requests will be in IANA considerations of the defining doc [17:55:45] next - suggestion that take text from TP OAM EXT draft and move here [17:56:08] will do, but do we need dest MEP-ID ? [17:56:30] Italo - need to understand use first [18:04:14] Yaakov suggested using FEC identifiers as in LDP [18:04:38] George says that Luca defined just IPv4 and IPv6, don't know of anything else needed right now [18:05:11] next issue - order of TLVs - the order should be handled in the protocol document, not here [18:07:47] next suggestion - alternative TLV structure with 2 bits - one gracefully ignore, and one if in authentication set [18:08:07] the best place to put these bits is the MSBs of the TLV type itself [18:09:35] Shahram - why not just authenticate the entire TLV set ? [18:16:19] decision - won't put in authentication bit - rather reserved bit [18:17:12] compacy null - the present null is rather large, instead of one byte [18:20:53] Yaakov at mike didn't like single byte from implementation point of view [18:24:47] done [18:24:57] are you sure [18:25:07] adrianfarrel leaves the room [18:25:08] YJS leaves the room [18:26:22] David Sinicrope leaves the room [18:35:58] Stewart Bryant leaves the room [20:20:39] Stewart Bryant joins the room [20:20:42] Stewart Bryant leaves the room [20:21:21] Stewart Bryant joins the room [22:08:22] Stewart Bryant leaves the room