Question

In a scenario where we have customers, vendors, & branches which all have common attributes such as an address, which is a better solution and why?

Solution 1:

• Customer table

• Vendor table

• Branch table

• Address table with nullable foreign keys for CustomerID, VendorID, BranchID

Solution 2:

• Entity table

• Customer table with EntityID

• Vendor table with EntityID

• Branch table with EntityID

• Address table with EntityID and a EntityType flag of ‘C’, ‘V’, or ‘B’

Solution 3 (suggested by AJC):

• Customer table

• Vendor table

• Branch table

• Address table

• CustomerAddress xref table between Customer and Address

• VendorAddress xref table between Vendor and Address

• BranchAddress xref table between Branch and Address

Solution 4 (suggested by 9000)

• Customer table

• Vendor table

• Branch table

• CustomerAddress with FK to Customer

• VendorAddress with FK to Vendor

• BranchAddress with FK to Branch

• vwAddress which UNION ALLs each of the above address tables, and includes a type flag (‘B’,’C’,’V’)

Notes:

Each customer, vendor, or branch could have multiple addresses, but should have at least one.

If an “entity” is both a customer and a vendor, it could have separate addresses for each role.

Would like to know if a customer is also a vendor.

Answer 1

C) None of the above.

The correct way to do it would be to have an Address Table and a CustomerAddress, VendorAddress and BranchAddress respectively to join each C V B entity with one or many addresses.

For attributes that are one-to-one, obviously there is no need for the extra tables, you just add the Id to the main table (AddressId)…