–redesign-per-request option for Path_Request

[akhil-mutike]

Hi Team,
I have observed that Path_Request has option called "–redesign-per-request" .Can some one give clarity what exactly this option does?
As far as I know there are two steps in network design:

1. Choosing the Correct Amplifiers depending on the span loses(this is independent of requests)
   2)Setting gain levels of amplifiers and launch powers of individual spans to get the best OSNR(this defiantly depend on Spectral Load)

1)Does Option give out a optimum scenario considering all the requests or the Worst case scenario?
2)Does this option also calculate the Optimum launch power for the reference span?

[EstherLerouzic]

Dear @akhil-mutike ,
The default design settings for the path-request-run script have been modified starting from v2.11. Before this release, the network was systematically redesign for each candidate in the service list, each request playing as a new candidate reference channel for the design. This behaviour led to non intuitive behaviour of the gnpy-path-request script, since the gain or delta_p of amplifiers could change at each request. After v2.11, it was decided to have the script behaviour be aligned with a standard planning tool: design is performed once for the reference channel defined in the SI block of the eqpt_config, and requests are propagated based on this design. The --redesign-per-request option can be used to restore previous behaviour (design using request channel types). This has been documented here https://github.com/Telecominfraproject/oopt-gnpy/blob/master/docs/release-notes.rst#v211.

Hope this helps!

[akhil-mutike]

Thanks Esther, for the clarification.

What exactly do you mean by gain or delta_p changes from request to request?
Gain is calculated based on span losses(so, independent of request). Coming to delta_p, It is just variable power added/subtracted based span loss with respect to reference span.

[EstherLerouzic]

I mean that GNPy systematically run an 'autodesign' function when running a simulation.

• If the imported network file already has a design (gain target or delta_p targets in amplifiers) and if the design is OK (meaning that values are consistent with max power of amplifier for example) the design is used as is.
• If the imported file has some inconsistencies issues or gaps in the design, then the function corrects or fill in the gaps.

In order to computes the targets, the algorithms computes the expected max power to be supported by amplifiers, and this max power depends on the channel types.

• Before release v2.11, the total power was recomputed using the channel type defined by each request to compute this max power. It means that the design could change between different requests for a same run, if the channel type changed the total power.
• After release v2.11, the design is no more recomputed at each request. Instead the design is made prior to all request run, using a reference channel type (the one defined by SI).

If a user wants the old behaviour back, he can use the option -redesign-per-request.

Hopes this clarifies!