It is your choice to define a descriptor or superdescriptor as unique. You can have more than one unique descriptor on a file (I don’t recommend it, but it is “legal” for Adabas). It is also legal (again I don’t recommend it) to allow null values for a unique descriptor. An ISN will always exist whether or not you use it directly and only one per record.
A unique descriptor can be alphanumeric, numeric or any valid Adabas format. An ISN is always an internal number assigned by Adabas. You can update the values, including unique descriptors, of a record, but you can’t update the ISN (of course, you can delete it and add it again).
So, yes, there is some overlap between the ISN and a unique descriptor, but one (ISN) is a system-managed value and the other is application managed. Either can uniquely identify a given record.