search

ietf-ccamp-wg / draft-ietf-ccamp-wdm-tunnel-yang

CCAMP WG repository for the ietf-flexi-grid-media-channel YANG model
Other
2 stars 2 forks source link

Review l0-path-constraints grouping #16

Closed italobusi closed 2 years ago

italobusi commented 3 years ago

Few questions:

  1. It seems identical to the wson-path-constraints grouping. Should it be moved to layer0-types-ext and imported instead?
  2. Should this grouping also needed to provide path constraints for tunnel setup?
danielkinguk commented 3 years ago

"1. It seems identical to the wson-path-constraints grouping. Should it be moved to layer0-types-ext and imported instead?" moved to Layer-0.

italobusi commented 3 years ago

2. Should this grouping also needed to provide path constraints for tunnel setup?

Still need to be addressed

sergiobelotti commented 3 years ago

The grouping is used in the request of tunnel input rpc, to provide the way to choose wavelength assignment e.g. randomly or taking into account residual capacity.

italobusi commented 2 years ago

@sergiobelotti : I agree that this attribute is useful for optical path computation

IMHO, if this attribute is an input for the path computation, it should be also provided in the TE tunnel model and not only in the path computation RPC, so the grouping should be used in other parts of the model and not only in the path computation RPC augmentation

Two doubts/questions:

  1. is this attribute defined/applicable per-tunnel or per-path? In other words, it is possible to apply different wavelength assignment mechanisms to the primary and secondary paths of the same tunnel?
  2. where are these mechanisms defined? The layer0-types-ext model references RFC 6163 but I am not able to find any definition of first-fit, random selection or residual capacity in RFC 6163 ...
danielkinguk commented 2 years ago

Italo and Sergio to discuss open issue next week, prior to our next call.

sergiobelotti commented 2 years ago

@sergiobelotti : I agree that this attribute is useful for optical path computation

IMHO, if this attribute is an input for the path computation, it should be also provided in the TE tunnel model and not only in the path computation RPC, so the grouping should be used in other parts of the model and not only in the path computation RPC augmentation

Two doubts/questions:

  1. is this attribute defined/applicable per-tunnel or per-path? In other words, it is possible to apply different wavelength assignment mechanisms to the primary and secondary paths of the same tunnel?
  2. where are these mechanisms defined? The layer0-types-ext model references RFC 6163 but I am not able to find any definition of first-fit, random selection or residual capacity in RFC 6163 ...

The wavelength-assignment method are just described by the identities defined in layer0-types-ext:

identity wavelength-assignment { description "Wavelength selection base"; reference "RFC6163:Framework for GMPLS and Path Computation Element (PCE) Control of Wavelength Switched Optical Networks (WSONs)"; }

identity unspecified-wavelength-assignment { base wavelength-assignment; description "No method specified"; }

identity first-fit-wavelength-assignment { base wavelength-assignment; description "All the available wavelengths are numbered, and this WA (Wavelength Assignment) method chooses the available wavelength with the lowest index"; }

identity random-wavelength-assignment { base wavelength-assignment; description "This WA method chooses an available wavelength randomly"; }

identity least-loaded-wavelength-assignment { base wavelength-assignment; description "This WA method selects the wavelength that has the largest residual capacity on the most loaded link along the route (in multi-fiber networks)"; }

italobusi commented 2 years ago

2021-11-29 Aihua/Italo/Sergio

The wavelength-assignment attribute should be defined for both path computation RPC and tunnel models.

The attribute is defined per-tunnel and therefore should be moved to the l0-tunnel-attributes grouping.

The definitions of the possible wavelength-assignment values is well known in literature but not defined in any standard document:

https://www.cs.sfu.ca/research/groups/NML/archive/pdf00002.pdf

This issue may be clarified in the layer0-types-ext text.

Since the wavelength-assignment attribute is optional, there is no need for the unspecified-wavelength-assignment identity. When the wavelength-assignment is not present no wavelength-assignment method is specified.

danielkinguk commented 2 years ago

Waiting on Sergio and Italo to update the model, then we can close the issue.