Transcript Relational Databases

Presentation by Irina Kogan.
November 27, 2000.
1
Background
 Business Object Notation (BON)
– object-oriented modeling language (classes and objects)
– inheritance, association, aggregation, constraints
expressed in class invariants
– two notations: textual and graphical
 Relational Databases
– What does a relation consist of?
• relational instance (table)
• relational schema describing the column heads for the table
– Entity-Relationship (ER) diagrams and SQL
• structural constraints, check constraints, assertions, triggers
2
Motivation
 Objects have identity, state, and behaviour in addition
to data.
 A relational database stores data only.
 In spite of these differences, a relational database as well
as BON have strong mechanisms for expressing
semantic constraints.
 As the importance of databases and people’s reliance on
them have grown, a mapping strategy built on a solid
understanding of the similarities and differences of these
models is more needed than ever.
3
Inheritance (BON -> ER)
:
4
Association (BON -> ER)
 3 possibilities for mapping associations:
• many-to-many
• many-to-one
• one-to-one relationships
 Example:
.
5
Mapping class invariants to
check constraints
 Example (uses the association diagrams on the
previous slide):
– Specify the following constraints (in BON and in SQL):
• a sailor’s rating must be an integer in the range 1 to 10
• no sailor can reserve a green boat
6
Mapping class invariants to
check constraints (cont’d)
Many-to-many relationship => 1 class -> 2 tables
Simple check constraints:
CREATE TABLE Sailors (sid INTEGER,
name CHAR(15),
rating INTEGER,
PRIMARY KEY(sid),
CHECK (rating >= 1 and rating <= 10))
Check constraints involving queries:
CREATE TABLE Reserves (sid INTEGER,
bid INTEGER,
day DATE,
PRIMARY KEY (sid, bid),
FOREIGN KEY (sid) REFERENCES Sailors,
FOREIGN KEY(bid) REFERENCES Boats,
CONSTRAINT NoGreenBoats
CHECK (‘green’ <> (SELECT B.color
FROM Boats B
WHERE B.bid = Reserves.bid)))
7
Additional mapping strategies
 Aggregation -> weak entities
 Clusters -> aggregation in ER-diagrams (no direct
correspondence in all cases)
 Specifying state change: use of the keyword old in BON,
use of triggers in SQL
 Mapping was also successful for:
• bidirectional links
• self references
 Note:
•
•
•
•
use object ids (OIDs) for primary keys
tables are persistent objects
void is similar to null (except for void references)
message passing can be modelled via triggers in some cases
8
Summary
 Many people believe that relational databases will
evolve in time and that the vast majority of
organizations will start mapping objects to
relational databases.
 This research has proven that most of semantic
constraints expressed in BON can be successfully
mapped to those in relational databases. The
additional work described in the literature read on
the subject has shown that objects themselves can
be mapped successfully.
9