Actually, I would not recommend using the PK's stored in the Hubs
or the Links
as Dimension and fact fillers. I would actually recommend you create your own Sequence numbers (new ones) for each dimension and each fact.
You do have to use the keys for the joins - that is true, but not for the output. You should be matching by Business Key for a type 2 dimension, then running a delta against the dimension.
With regard to the actual time-line, you need to decide (by user rules) which temporality / which time line to use. In other words, there may be more than just the original LOAD and LOAD END date available. There may be an estimated date, or an actual date, or a projected date, or a booked-date, or some other business date. In reality, your queries should focus on using BUSINESS dates for both query alignment (sorting and joining), AND for delta processing on the way to the type 2 dimension.
Hope this helps,