100 likes | 233 Views
Identifiers for MPLS-TP. George Swallow. Status. In Anaheim, draft was identified as a pre- req for G.8110.1 Incomplete draft was rushed to last call Changes in various Framework drafts as well as a volcano made it impossible to produce a quality draft in time for ITU meeting
E N D
Identifiers for MPLS-TP George Swallow
Status • In Anaheim, draft was identified as a pre-req for G.8110.1 • Incomplete draft was rushed to last call • Changes in various Framework drafts as well as a volcano made it impossible to produce a quality draft in time for ITU meeting • Now declaring that the draft is NOT in last call
Changes in rev -02 • Incorporated most of the ITU liaison comments • Addressed many of the mail list issues • Much rewriting for clarification • Many comments have not been addressed
Clarifications • Equated Interface (IF) to Access Point (AP) • Clarified use of source and destination • via notation in BNF • Noting that in a configured environment East and West would be equivalent • Src/Dst terminology is inherent in GMPLS
MPLS-TP LSP Identifiers LSP-num – 16 bit identifier as in RFC3209 Unique within scope of tunnel LSP-ID formed as local{Global-ID::node::Tun-ID} +remote{[Sp-ID]::node::Tun-ID}::LSP-ID • Canonical Format of LSP-ID • lower ([Sp-ID]::Node-ID) goes first • Compatible with GMPLS signaling
Rationale for two Tunnel-IDs • Primary motivation is to have a simple way of forming globally unique MEP-IDs • [Global-ID]::Node_ID::Tunnel_ID::LSP_ID • Similar to Pseudowire Attachment Circuit ID • It has also been pointed out that this will be useful for signaling associated bi-directional lsp
Example of Associated Bidirectional LSP • Node A signals tunnel 5 to node B with an object requesting an associated bidirectional tunnel • Node B signals tunnel 7 to node B returning the object completing the association • This also raises the question of whether two LSP-IDs are needed • (I don’t think this is necessary for initial setup, but for regrooming it could be an issue)
Pseudowire Maintenance PointsTentative resolution +-------+ +-------+ +-------+ +-------+ | | | | | | | | | A|---------|B C|---------|D E|---------|F | | T-PE1| | S-PE2| | S-PE3 | | T-PE4 | +-------+ +-------+ +-------+ +-------+ The identification for the Pseudowire is: AGI = AGI1 Src-Global_ID = GID1 Src-Node_ID = T-PE1 Src-AC_ID = AC1 Dst-Global_ID = GID1 Dst-Node_ID = T-PE1 Dst-AC_ID = AC4 MEP_ID at point A = AGI1::GID1:T-PE1::AC1 MIPs same LSPs PW status from a S-PE is sent by the node
Service Provider IDs Current Status: Two formats Global-ID as per RFC5003 ITU Carrier Code Issue: Should these be combinable with all other identifiers that need global uniqueness Or should some limits exist on mixing and matching ITU and IP style IDs? Need to figure out exactly where these will get used before deciding
Other open Issues • Many open issues – this is a partial list • IPv6 • Support for all forms of IF-IDs allowed in GMPLS • Support for other PHOP formats for identifying Ifs • Two LSP-IDs as well as two tunnel-IDs • Do we need circuit-IDs as well as MEG-IDs