星型图：事实与维度之间的三重关系

Question

I'm modelling a Star Schema from an ERD. The Database is a rental firm. My fact table now contains a single Rental Booking. Customers are able to book cars, collect cars and return cars. The date of when this happens should be recorded.

So, my Star Schema now has a Time Dimension with Day/Month/Year and three relations to the fact table. I'm wondering if that's the right way to do or if there are other options to consider, because I have never seen a Star Scheme where a dimension has more than one Relation.

This is it:

 Booking(Fact)
 bookingId
 bookingDate(FK)
 collectionDate(FK)
 returnDate(FK)

 Time(Dim)
 timeId
 day
 month
 year

Answer 1

Yes, you can have a Fact join to the same dimension multiple times. I've seen this used with Time dimensions especially. I'm not sure what key you are using to join the Fact to the Time dim, but you may end up with a query like this, when selecting the data:

select bFact.bookingID,t1.day "Booking Date",t2.day "Collection Date",t3.day "Return Date"
from Booking bFact
join Time t1 on t1.date = bFact.bookingDate
join Time t2 on t2.date = bFact.collectionDate
join Time t3 on t3.date = bFact.returnDate

Answer 2

This is a diagram to try and help you understand better.

If your bookingdate, collectiondate, returndate are all part of the time dimension then you dont need them in the fact table. They need to be in the dimension table.

I have created a diagram that would correctly demonstrate what i think you mean. As an example, if return can have different measures then it needs to be a separate dimension. Each dimension will always need a primary key and these will always correspond to a foreign key in the fact table.

Any questions let me know