Understanding Virtual Feeds

When do you need to make use of a virtual feed vs doing processing with input and storing the value in a feed?
I would like to calculate when solar generation goes above 2.5kW. I can program this with processing from the input, so that a 1 is stored in the feed. Is this a correct implementation? Or should I make use of a virtual feed? What are typical use cases for virtual feeds? Could data of virtual feeds be extracted in an automated way to be used in another application?

Here’s a read on virtual feeds that helped me. One disadvantage of virtual feeds is you cannot log a virtual feed, which requires me to create a fake input and do the processing that way in order to log it. Virtual Feeds relies on a source feed that you already are logging from your processing input(s)

In my case. I have Amps for Phase A and Phase B in my house, and in a virtual feed, I select the source feed PhaseA and add a feed Phase B, it just adds the values together or what ever processing you want to do, and it then can be added to a dashboard or graph.