I have a perhaps slightly-unconventional way that I like to handle problems like this, which throws back to how you’d (have to) do it with a database.   Instead of building a lot of “references” from one Perl object to another, I arrange for the various objects to descend from a common parent-class which gives each object a short random-string “moniker” by which any object-instance can be uniquely known.   It is, in effect, a primary-key.   I might then store the objects in a hash by moniker.   I don’t fool around with stuff like use-parent.   I tackle the requirement in-memory exactly as I would have to tackle it in a database schema.