Hi Jonas. You need to consider that what is in the MSA spreadsheet Local control of OpenROADM describes the behavior of local controller (embeded in the device). What we implement in transportPCE is compliant whith what is described in the white paper (focusing on actions to be performed by the ROADM network controller). So we send commands through netconf, and locally, devices controller interpret this command and trig local actions. As the output power set for an add channel is corresponding to the output of the line booster, my understanding is that once WSS and internal attenuators have been set, we should go back to gain control mode for standard operation for this amplifier (which also support channels that are optically bypassed in the node : express channels).

I understand the distinction between a local embedded controller and an SDN controller (such as TransportPCE). In my understanding it is the task of the SDN controller to coordinate and configure the use of gain/loss vs. power mode (per channel) for all local controllers. Gain/loss mode refers to all components in the path, not only amplifiers but also WSS, VOA, etc. So the booster amp can operate with constant gain and all express channels be in gain/loss mode while at the same time keeping add channels in power mode by e.g. having the local controller continuously controlling the add channel attenuation in the WSS. I think this is what the MSA refers to when it says "the Add/Drop is likely to stay in Power mode". The motivation for switching local controllers to gain/loss mode is to avoid the potential instability of having multiple power controller compensate for a change in loss at the same time. As described in the MSA:

"The SDN controller will then be able to sense the drop of power and localize the fault to the span where it occurred.  It can then adjust the gain of the node immediately down stream from the fault and all other nodes will be at their appropriate target already."

In the case of an add channel, if a fault occurs in the node where it is added (e.g. a pinched patch cord), this should always be compensated by adjusting the gain in the same node, which will happen faster if the channel is in power mode. A fault downstream of the node where the channel is added will have no impact on the power control so there is no risk of instability in this case.

OK. Understand your point. What is in the MSA has been specified at the begining of OpenROADM. I have asked several time for clarification on it (since I did not have your level of comprehension...), but did not get a clear answer about it. What you explain make sense (I did not realize that in case of an add, mode was applying to control loop for WSS and VOA. I thought the amplifier was included in the loop. However, what we do is what is specified in the WP. If I remember correctly, for add and express channels, set the mode to gain/loss at the end, whereas for drop, let it to power. What I suggest is that you raise an issue explaining your point as you did it here on https://github.com/OpenROADM/OpenROADM_MSA_Public/issues : this is a public site where anybody can raise an issue about OpenROADM. Because at the end, only the people that were involved in the description of this process will have the right answer and can confirm you whether this is feasible or not.