Newcomer feedback from lattelecom

[ggrammel]

In TIP Workplace managed to find Polito model presentation and as I understood actual code is using these principles. This kind of info, at least for me, is helpful to understand the model.

As not professional coder and with no experience with open-source projects I don’t understand the principles how new functions are added (organizational task) – someone adds something and then hopes for pull approvement or there are some guidelines what functions project is missing at the moment and these are pursued with high priority (something besides Raman model).

On fresh Ubuntu 18.04 installation, as i remember, besides requirements.txt python3.6-tk installation was needed. Readme doesn’t mention that this library is using python 3.6, only the onboarding guide. In Ubuntu 16.04 with default python 3.5 additional repository is needed for ‘apt install python3.6’ to work.

Results output in console are understandable, as was mentioned in the call JSON result output would be good idea - simplify gnpy integration somewhere.

Already opened a few issues: · ROADM loss o After looking in Polito model presentation it is much clearer how ROADM model can look in the future o If for some purposes channel drop power calculation would be needed, then losses should be divided in three parts – add_loss, drop_loss, express_loss. Calculation of these for network.json can be made by front end o For low loss spans only preamplifier can be needed (no booster). At the moment all hops are automatically calculated with Preamp and Booster

· EDFA parameters and booster-less type o Introduction of Booster-less amplifiers o Mentioned Gain calculation from ‘Power/channel@output – Power/channel@input’ is just like our vendor equipment is configured and can be recalculated for Gain parameter by front end

In general, two modes of operation would be needed: · Automatic – auto calculation of better Amp and so on · Manual/Manual-Auto – calculation with manually entered parameters (Amp, fiber, span padding, …), that is helpful if performance calculation is made including already commisioned network equipment (old network + new network or ROADM direction), sometimes only particular card types are better choice (spare parts, end of sale)

[dutc]

@jeanluc-auge Can you comment?

[jeanluc-auge]

1. About 2 modes of operation: manual vs automatic for now there is no toggle between the 2 modes, though they are supported:
   • if amplifiers are inserted after each fiber and ROADM elements, with non 0 gain parameter, the operation is manual
   • if amplifiers are inserted but some or all of them have 0 or no gain parameter, the tool will set the proper gain to compensate for previous element loss.
   • if some amplifiers are inserted but not all, the tool will complete automatically the missing amps

This way the tool autodiscovers what needs to be completed. However, as a result, the low span loss scenario that you describe where you skip the egress amp at a ROADM location is not supported. Because the tool would automatically add the missing amplifier. The only way to support it would add a toogle automatic / manual design. This can be discussed but I am personally not in favor:

• added management complexity
• if manual design is selected and some configurations are missing, you'll get an error message instead of seamless completion

• and the main reason for me: this scenario is a metro use case, in long haul you wold not skip a roadm egress amp and prefer padd the span or fuse it with the following one. For performance reasons (considering roadm losses) and because some power roadm control loops require an egress amplifier.

  2. about ROADM loss I have implemented an optional power mode in the tool, which sets the ROADM egress power target regardless of express or add origin. Even if you do a power sweep, you will always have the same power exiting the ROADM, which is more realistic than a fixed roadm loss because it reproduces what a real equipment power control loop does. It is parametrized in the eqpt_config.json:

     "Roadms":[{
         "gain_mode_default_loss": 20,
         "power_mode_pref": -20
         }]

     ad the power mode configuration:

     "Spans":[{
         "power_mode": false,

     Let me know what you think of it.

---

3. about automatic design Now that the power mode is implemented, I will be working on autodesign with different power settings in the spans. Said differently, the gain of the amplifier would not just compensate the previous span loss but could be lower or greater depending on the power difference between the 2 spans.

[berahtlv]

1. Ok.
2. I like it, interesting idea with SI.pref As I understood SI, Roadms and Spans in equipment.json are helpers. In my opinion equipment.json should contain only fixed equipment parameters for particular type and subtype and connector or roadm or some other that type losses or parameters are just for particular instance. "gain_mode_default_loss" (duplicates "loss") and "power_mode_pref" should be in element "params" and can be optional. "power_mode" is more useful as console parameter, then calculation method can be switched as required.
3. Ok.