search

IETF-TEAS-WG / actn-poi

Applicability of Abstraction and Control of Traffic Engineered Networks (ACTN) to Packet Optical Integration (POI)
Other
5 stars 2 forks source link

Muxponders' constraints #108

Closed italobusi closed 12 months ago

italobusi commented 1 year ago

If the muxponders can constraint which client ports can be connected together, this constraint shall be reported to the MDSC through the MPI but this information is not defined in the existing topology YANG models (it could be gap).

References:

italobusi commented 1 year ago

2022-11-29 ACTN POI Call (Slot 2)

italobusi commented 1 year ago

2022-12-13 ACTN POI Call (Slot 2)

  • [ ] @italobusi @sergiobelotti @jbouqui153 @FabioPeruzzini check whether muxponders can constraint which client ports can be connected together (talk offline also with Gabriele)

Jeff checked internally and there are muxponders which have this constraint but other muxponders which do not have this constraint

Italo: my doubt is about how to report at the MPI whether there is or not this constraint in the underlay hardware

Sergio: this is true for any other hardware constraint

Jeff: you can also have a muxponder on one end and a transponder on the other end. You can also have multiple generations of transponders/muxponders that are connected as long as there are common operational modes between the older and newer transponder/muxponder generations

Italo: I think this is related to the open issue of reporting the L1 capabilities/constraints of a transponder

Italo: we may have an high level description of this issue in the appendix and leave the detailed analysis and resolution to other documents

Jeff: let's further discuss with Gabriele

ggalimba56 commented 1 year ago

Basically the “old” muxsponders (e.g. 10 x 10GE —> 100G) have have a fixed mapping between ports and 100G time slots, so port 1 is always connected to TS1 etc. This is clearly a constraint. But more recent MXP or if you want crossponders are flexible and the mapping Port <—> time slot is provisionable. In summary, we need to support the constraints disclosure to MDSC for old MXP but also we need to support the possibility to provision the MXP switch matrix. What if we keep the switch matrix description and we add a flag saying provisionable or fixed ? If provisionable will be r/w if fixed will be r only (reporting the port/TS mapping)

italobusi commented 1 year ago

2023-01-10 ACTN POI Call (Slot 2)

  • [x] @italobusi @sergiobelotti @jbouqui153 @FabioPeruzzini check whether muxponders can constraint which client ports can be connected together (talk offline also with Gabriele)

It is confirmed that this is a possible scenario to be addressed. The draft can just describe the issue and identify this as a gap of the current models. The gap can be then addressed in the context of optical impairments and/or flexi-grid topology and tunnel models.

italobusi commented 1 year ago

2023-05-09 ACTN POI Call (Slot 2)

It has been agreed to update the draft to just describe the issue and identify this as a gap of the current models.