Entity Relationship Modeling (ER modeling) is by far the most common way to express the analytical result of an early stage in the construction of a new database. In this ebook, Alf Pedersen describes the principles for ER modeling, as well as the most important terms used in modeling a new database.
Do they seem a bit far-fetched? Yes they are. They are a signal of lacking analysis, for all practical purposes. Most likely, as said, the supplier product entity is really nothing more than a set of attributes in the product entity.
One-to-one relationships are a signal that you have different entities that are probably the same entity.
One-to-many relationships
Mandatory-mandatory
This signals that an order (in this case) MUST belong to ONE AND ONLY ONE customer, and a customer MUST place ONE OR MORE orders. While theoretically possible to construct, this relationship would violate constraints in the database: It will not be possible to create a new customer if there are no orders; however, an order cannot be created without customers.
Mandatory-optional
A new customer MUST place one or more orders, while an order MAY belong to a customer. Note that the order can exist without a customer, while you cannot have customers who do not have any orders. Well, there might be such a situation, but I find it hard to believe. I would have looked more deeply into the problem domain. This relationship is possible, but relates to odd business rules.
Development Cycles 
/ 74 
Enterprise Java and Rational Rose -- Part I
