IETF 97 MPLS WG draftietfmplsrmr Kireeti Kompella Luis

IETF 97 -- MPLS WG draft-ietf-mpls-rmr Kireeti Kompella & Luis Contreras

Status • Draft is progressing well • On-going technical discussions • Some very early implementation experience • Big change: link-centric to neighbor-centric • IGP TLV for auto-discovery has changed to reflect this • Auto-discovery procedure has been tweaked, but details are still being worked on

Ring Relationships AC 0 These relationships used to be wrt ring links. 1 CW Express 2 3 It makes more sense to define the relationships wrt nodes: 1. Using nodes eliminates the possibility of contradictory relationships in the case of multiple links between nodes 2. It also means that new links coming up between nodes will automatically be included with the right relationship 3. It also means that links don’t have to be identified or exchanged across the ring 4. Finally, the new encoding is much more efficient

IGP TLV changes • Previously, the IGP had two top-level TLVs: • the Ring Node TLV and the Ring Link TLV • The Ring Link TLV identified a link by interface indexes or interface addresses (local and remote) and listed all the rings going over that link

Old Ring Link TLV (IS-IS encoding) 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type (TBD) | Length=8+6*N | My Interface Index. . . | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |. . . (continued, 4 octets) | Remote Interface Index. . . | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |. . . (continued, 4 octets) | Ring ID 1 (4 octets). . . | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |. . . (RID continued) | Flags (2 octets) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Ring ID 2 (4 octets) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Flags (2 octets) | Ring ID 2 (4 octets). . . | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |. . . (RID continued) | Flags (2 octets) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |. . . etc. | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

New Node TLV (one per ring) 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type (TBD) | Length = 6+S | Ring ID (4 octets). . . | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |. . . (RID continued) | Node Flags (2 octets) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | sub-TLVs, if any. . . +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ S is the total size of the sub-TLVs

New Neighbor Sub-TLV (one per Node TLV, containing all neighbors) 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Type (TBD) | Length = n*6 | Neighbor Loopback. . . | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |. . . (continued, 4 octets) | Neighbor Flags (2 octets) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Neighbor Loopback (4 octets) | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Neighbor Flags (2 octets) | (etc. ) +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ n = number of neighbors included in the sub-TLV

New draft for IS-IS and OSPF encodings • The RMR draft now only has the high-level view of the TLV and sub -TLVs • The details have been moved to a combined IS-IS/OSPF draft • This is so that those WGs can discuss and comment on the encodings • The details of how auto-discovery works will remain in the RMR draft, as these are independent of IGP operation

New IGP-independent TLV Format (RMR draft) TLV: [RMR Node Type][RMR Node Length][RID][Node Flags][sub-TLVs] Sub-TLV: [RMR Nbr Type][RMR Nbr Length][Nbr Address][Nbr Flags]