IETF 77: Open Shortest Path First (OSPF) WG Agenda Friday, March 26, 2010. 1300-1515 Afternoon Session I/II Huntington Room ======================================================== Chairs: Abhay Roy Acee Lindem WG Status Web Page: http://tools.ietf.org/wg/ospf/ 1) Administrivia - blue sheets - scribe/jabber - Jabber room: ospf@jabber.ietf.org 5 minutes No keg of beer due to lack of donations. :-) 2) WG Status, Update 20 minutes Published since Stockholm RFC 5613, 5642, 5643, 5709, 5786, 5820 ospf-af-alt-10 in RFC editor queue mt-mib is stalled 3) Multi-Topology/Multi-Instance OSPFv3 for IPv4-Embedded IPv6 - http://www.ietf.org/id/draft-boucadair-ospf-v4v6-ospfv3-mt-02.txt - Dean Cheng 20 minutes Requesting MTID/MI ID for IPv4 embedded in IPv6 Similar proposal in IS-IS - where local configuration of MTIDs proposed. If same done in OSPF then this draft is not needed. Two operators have agreed that local configuration is OK. Abhay: IS-IS proposal for MT - similar needs to be done for MT OSPFv2 Dean: Proposal is for IPv6 (OSPFv3) which is not published yet - so ID allocation changes can be updated before publishing 4) OSPFv2 Router-Bit (R-Bit) - http://www.ietf.org/id/draft-pillay-esnault-ospf-rbit-00.txt - Padma Pillay-Esnault 20 minutes Capability to avoid using node as transit node exists in OSPFv3 - this is to add same cpapability to OSPFv2. Define R-bit to say no transit (when clear) Define S bit to indicate that the R-bit is supported (when set) Assigns to unused bits in LSA header Transit links set to LSInfinity to reduce probablity of blackhole in transition cases. Modifies SPF to skip no transit nodees Wenhu How does R bit differ from OL bit in IS-IS? Abhay: They are the same Wenhu: Why does S bit matter? Michael: S bit only affects how SPF is calculated Padma: S bit reduces churn and negative effects. Reduces window Dmitri: Could this be put in router info LSA - at least in first stage? Michael: This could delay when condition is detected and introduce longer period of blackholing. Padma: Solution tries to minimize disruption. Acee: Concerned about using 25% of available bits to solve problem addressed by maxmetric. S bit belongs in capabilities. Capablities LSA requires correlation so would be safe. Padma: Different problem than maxmetric Acee: Then don't need backwards compatibility - maxmetric has been sufficient for many years Padma: Wants to exclude the node Dmitri: Prioritize router info LSA to make it processed more quickly Padma: Welcome the feedback - may try to solve the problem differently. Abhay: Problem worth solving - but prefer different use of bits. Acee: Most people feel problem is useful to solve - take to list. 5) OSPFv3 PE-CE Routing Updates - http://www.ietf.org/id/draft-ietf-l3vpn-ospfv3-pece-05.txt - Padma Pillay-Esnault 20 minutes Last call ended in L3VPN in 4/09 but last call not completed in OSPF. In responding to comments simplified the solution: Rely on Domain ID only (no longer use Instance ID) Acee: Extended community ID had similar comment Padma: Semantics of community different for V2/V3 Acee: Ready to take to LC when authors are ready Acee: Increased amount of IPv6/V3 deployments - timely to have this published 6) ASON Update/CCAMP Discussion Summary - http://www.ietf.org/id/draft-ong-gmpls-ason-routing-exper-01.txt - Lyndon Ong 20 minutes Also presented at CCAMP Experimental status of IETF ASON drafts is hindering OIF - wants them to go to standrads track. OIF has been doing interoperability testing. Review differences between OIF testing and IETF specs. Acee: TDM bandwidth is not the same information in both cases Lyndon: Proposal to address that but concern about the amount of info needed to send all the possible information. CCAMP is interested in doing 5787bis Acee: Nice to have commonality between MPLS-TE and GMPLS Lyndon:Added varying prefix length for ASON requirements. Lou: ISID - bis version likely to change to proposed standards. If there are substantive changes, these need to be shown as necessary. Can't drop GMPLS functions Dmitri: With multi instance, accounting of bandwidth can be done in other ways - not precluded. Lou: Yes - not precluded but should not be part of base documents Lyndon: No multi-layer testing Lou: Interoperable implementations are the focus 7) OSPF Extensions in support of O-E-O pools in GMPLS controlled all-optical networks - http://www.ietf.org/id/draft-peloso-ccamp-wson-ospf-oeo-01.txt - Giovanni Martinelli 10 minutes Acee: GMPLS pushed to CCAMP while busy w MANET - now we have time to work on GMPLS in OSPF (like IS-IS) Need to address spatial/spectral assignment of LSPs - not just bandwidth Details presented on proposed LSAs information to be advertised Acee: This is not link information - so need new TLVs. May also want to separate dynamic from static. Unknown: What kind of nodes can be represented? Will it describe the types of connectivity which are possible? Giovanni: Yes - this is covered. Acee: May progress in OSPF even if needs CCAMP oversight as well. This is at transport layer so scaling is less of an issue than it is at L3. To be followed up after this meeting: 1)R-bit discussion 2)Publish MT MIB (end stalling) - not that much new/controversial even though no implementation (as experimental) 3)WG LC V3 PE-CE draft Whether will meet at Maastricht is TBD (probably determined in May) Adjourned