Draft-mpls-tp-OAM-maintnance-points-00 11th Nov. 2009 Hiroshima Yoshinori Koike / NTT

Background *Carrier Grade packet transport network requires MPLS-TP OAM function to be effective in maintenance operations *If some faults or un-expected events occur, quick actions and methods are mandatory in terms of transport network requirements.

Motivation of this I-D *Identifying the location of OAM maintenance points (MIPs and MEPs) is the most significant factor in MPLS-TP OAM framework. *How and where they are specified must be specified clearly somewhere in the MPLS-TP framework document. Why? If OAM maintenance points are not appropriately positioned, -the OAM function may be meaningless, however excellent those OAM functions are. -We cannot take appropriate action and it takes a lot of time to solve a problem.

Scope and objectives Clarify requirements regarding the location of maintenance points (MPs), that is, positions in MPLS-TP network and Network Element (NE) where OAM MPs need to be set Describe why these MPLS-TP OAM maintenance points are requested at those positions

Potential problem and solution concerning MPs Data flow Type 1) MPLS Control Msg NE NE Control packets are generated/received at one point somewhere within an NE (Not clearly specified) NE NE Type 2) MPLS-TP OAM Msg Ingress Egress Ingress Egress Regarding OAM packets, ingress and egress distinction is requested for effective maintenance operations (Points of interface, one on each side of switching fabric) Control packets are generated from one point somewhere within a NE

Typical maintenance operations *Quick location of fault point Scenario 1) -Within one administrative domain or outside the domain? Scenario 2) -Within a node or a section (between two neighbor NEs)? *Prompt handling of the problem -Replacing fault package(s) -Fixing or changing configuration in package, port, section, logical path and so on.

Scenario 1 (within administrative domain or not) Type1) Operator’s Administrative domain SN1(NE1) SN2(NE2) SN3(NE3) Customer NE1 Customer NE2 P1 P2 P5 ? ? ? Customer domain Customer domain PW/LSP Packet Loss Packet Loss MEP1 MIP1 MEP2 OK Type2) Customer NE1 Customer NE2 P1 P2 P3 P4 P5 P6 OK ? ? Customer domain Customer domain PW/LSP MIP MEP Legends Physical Interface Switch Fabric MEP1 MIP1 MIP2 MIP3 MIP4 MEP2 Loopback OK (1) (2) OK

Scenario 2 (Within a node or section) Type1) Operator’s Administrative domain SN1(NE1) SN2(NE2) SN3(NE3) Customer NE1 Customer NE2 P1 P2 P5 ? ? ? Customer domain Customer domain PW/LSP Packet Loss Packet Loss MEP1 MIP1 MEP2 NG Type2) Customer NE1 Customer NE2 P1 P2 P3 P4 P5 P6 OK ? ? Customer domain Customer domain PW/LSP MIP MEP Legends Physical Interface Switch Fabric MEP1 MIP1 MIP2 MIP3 MIP4 MEP2 Loopback OK (1) (2) NG

Requirements 1) At the intermediate node on PW/LSP, it should be possible to activate two MIPs, one on each side of the switch fabric/forwarding part. 2) At the edge node on PW/LSP, it should be possible to activate one MIP and one MEP, one on each side of the switch fabric/forwarding part. 3) If OAM is supported, MEPs have to be active at all times when the LSP is set, an MIP only has to be activated when it is necessary or required (e.g. for fault location). (That is, an MIP does not need to be active at all times.)

Large factor in packet delay or packet loss Next Steps Planning to work further on the text -Add other OAM requirements and their backgrounds (ex. Delay measurement, Packet Loss measurement) -Clarification of the meaning of “Loop back” LBM/LBR or traffic loopback? Loop back as a tool or as a function? -Requirements need to be developed and modified Distinction between purpose and means, Additional terminology (Ingress and egress) Soliciting comments, feedback and suggestions for this draft from WG Type1) Large factor in packet delay or packet loss Type2)

Thank you