Versions Compared


  • This line was added.
  • This line was removed.
  • Formatting was changed.
Comment: Migrated to Confluence 5.3


24 October 2017



  • Record
  • Review Agenda
  • Review FRD
  • Review F2F Agenda
  • Review CORD Build slides
  • Any other business

Discussion Items

20Review of Draft FRDLyndon
  • see onf2017.500 for draft FRD Contents
  • Review open items
  • Multi-domain/Multi-layer example (see next item below)
  • Any objection to liaison to OIF? Note that multilayer example is still under discussion. State clearly it is draft text, SDK normative
20Multi-layer ExampleKarthik/Lyndon
  • Note onf2017.041 identifies possible additional actions when server layer connection is created, esp. creation of new SIPs
  • Karthik's comments:
    • we have been over this before. I do not believe it necessary to create SIPs dynamically although TAPI does not prevent it. The options are
      - do not advertise SIPs at the internal layer transition points (base TAPI use case). INNI are created only at vendor or admin domains. This is also how MEF is approaching the problem. Check with Andrea for c-tag in s-tag case which is similar
      - advertise a multilayer SIP at the layer transition that maps to both NEPs (ETH & ODU)
      -  advertise 2 single layer SIPs
    • Discuss further at f2f

The following related snippet from previous TAPI interim meeting notes (onf2017.117)

Multilayer Issues

Reference document onf2017.041

FRS Multi-layer options:

  • Issue of whether you create one multilayer service or two single layer service
  • Option A (multi-layer topology with Transitional Link)
  • Option B (multiple single-layer topology with multi-layer SIP)
    • Replace Transitional Link with Service Interface Point
      • Seems to be a change to make SIP one-to-one with NEP
      • Need mapping between client link and server conn
  • Clarification: the decision whether to export a single multi-layer topology or multiple single-layer topologies is taken by the Domain Controller. The MD controller shall be capable to deal with all the variations.

Client Layer Mapping Issues

  • Open Issue: How to efficiently model Fixed/limited-flexibility mapping capability?
  • How does the Domain-1-Controller know that it has to create an ETH Link after having created an ODU2 connection (with option A)
  • How SEPs can be dynamically created (for option A)
  • How to model TL in case multiple client layers can be supported by the same server layer

Sub-layer model (e.g. HO/LO ODU, mac-in-mac) Issues

  • Open Issue: How to model hierarchy within the same layer (e.g., LO-ODU multiplexing into HO-ODU) especially when flexibility is possible at both sub-layers: TL between NEPs on the same Node?

Multi-Domain stitching

    • One option is to use common Service Interface Point Naming
    • Open Issue: How the MD controller can glue together the topologies exported by the Domain Controllers into a single topology (GitHub open issue #190)


Review of F2F AgendaNigel?
  • TAPI scheduled for mornings Mon., Tues., Thu. and Fri.?
  • Anyone still needing to register for Mon/Fri sessions? Registered Karthik - Registration for Monday & Friday
  • Review of oimt2017.KL.001.01 on OIMT wiki
  • Topics: added to KL.001
    • FRD Review
    • OT IM Spec Models
    • MEF Topics? Karthik - MEF has asked for bidirectional associations - Nigel: will not be in core - need further discussion with MEF? Puts significant load on implementations
    • Ask Stephane St-Laurent for update on TIP work? Yes, if he is willing
20Review of CORD Build SlidesLyndon
  • see otcc2017.LYO.001 - slides need further work
  • Note: Karthik/Guru discussed potential ONOS/TAPI Brigade to write driver for ONOS but need to identify someone with software background who could lead the effort
20CORD Science FairLyndon
  • Contacted Ricard about Ref Impl - can we show this? Use laptop
  • Received link from Sedona for video from OIF 2016 Demo + offer to present it

Old Action Items

    •  All: Consider what needs to be covered in the TAPI portion of the interim face to face meeting
    •  All: Propose approach to demonstrating TAPI at CORD Build event
    •  Nigel Davis: Validate with David that we do have two booths at the CORD Build Science Fair 
    •  All: Review the FRD by 24 Oct 2017
    •  Karthik Sethuraman: add slot to TAPI calls to cover FRD.

New Action Items

    •  All: location to post recording that is accessible to China - maybe mirror
    •  Karthik, Lyndon: update draft FRD example for Friday liaison to OIF
    •  Lyndon: check with Guru on availability for TAPI review
    •  Lyndon: Get David to change agenda title for morning CORD session
    •  All: continue to review FRD
    •  Nigel, Karthik, Lyndon and anyone else interested: continue to work on CORD Build slide pack and Science Fair
    •  Nigel: host meeting next week (Karthik may be on plane, Lyndon in Shanghai)