If there is only one child, you can use the well-known key name so that you can directly extract the child (since you know the parent key). This can be useful, because at any time when you know the name or identifier of the parent key, you can easily get the child directly.
child_key = db.Key.from_path('ChildKind', 'knownname', parent=parent_key) child = db.get(child_key)
If there are several children, you can potentially save the list of keys (it can be unindexed) on the parent object or use the request of the ancestor. I would say it depends on how your data is updated, communicated (presented) and the amount of data. In other words, it depends on the needs of your application.