Information Systems Education Journal (ISEDJ) 9 (4)

September 2011

Teaching Case

Solving Relational Database Problems with

ORDBMS in an Advanced Database Course

Ming Wang
ming.wang@calstatela.edu
Department of Information Systems
California State University
Los Angeles, CA 90032

Abstract

This paper introduces how to use the object-relational database management system (ORDBMS) to
solve relational database (RDB) problems in an advanced database course. The purpose of the paper
is to provide a guideline for database instructors who desire to incorporate the ORDB technology in
their traditional database courses. The paper presents how to use the specific object-relational
database (ORDB) technology to solve three normalization problems: Transitive dependency, Multi-
value attributes, and Non-1st Normal Form. The paper also provides the solutions to data complexity
problems with three specific ORDBMS techniques: object view, object inheritance, and object
integration. The paper summarizes the significance and advantages of teaching ORDBMSs in advanced
database courses. Course contents and students’ learning outcomes are discussed. To be more helpful
to database educators, the paper presents a complete object-relational database development case
study from the UML class diagram design to Oracle ORDBMS implementation.

Keywords: Object-relational database, Database Curriculum, Oracle Database, Normalization

1. INTRODUCTION into the commercial database market has caused

the database professional’s attention in seeking
The success of relational database management how to utilize its object-oriented features in the
systems (RDBMSs) cannot be denied, but they database development and has brought new
experience difficulty when confronted with the challenges for IS instructors in teaching
kinds of "complex data" found in advanced ORDBMS in their database courses. In response
application areas such as hardware and software to this challenge, the author has incorporated
design, science and medicine, and mechanical the ORDB technology into her advanced
and electrical engineering. To meet the database course. ORDBMS enhances object-
challenges, Oracle, IBM and Microsoft have oriented technology into the relational database
moved to incorporate object-oriented database management system (RDBMS) and extends
features into their relational DBMSs under the traditional RDBMS with object-oriented features.
name of object-relational DBMSs. The major As an evolutionary technology, ORDBMS allows
database vendors presently support object- users to take advantages of reuse features in
relational data model, a data model that object-oriented technology, to map objects into
combines features of the object-oriented model relations and to maintain a consistent data
and relational model (Silberschatz, et al., 2009). structure in the existing RDBMS.
The emergence of object-relational technology

©2011 EDSIG (Education Special Interest Group of the AITP) Page 80

www.aitp-edsig.org /www.isedj.org
Information Systems Education Journal (ISEDJ) 9 (4)
September 2011

The purpose of the paper is to provide a guide limited inroads during the 1990’s, but have since
for database instructors who desire to been dying off. Instead of a migration from
incorporate the ORDB technology in their relational to object-oriented systems, as was
traditional database courses. This paper widely predicted around 1990, the vendors of
presents how to use ORDBMS to overcome relational systems have incorporated many
relational database weaknesses and solve some object-oriented database features into their
existing normalization problems. The paper first DBMS products. As a result, many DBMS
introduces the background and features of products that used to be called “relational” are
ORDBMS, then presents how to use the specific now called “object-relational.” (Garcia-Molina,
ORDBMS techniques to solve normalization et al. 2003).
problems in 1) Transitive dependency, 2) Multi-
value attributes, 3) and Non-1st Normal Form, Practically, ORDBMS bridges the gap between
and how to use the specific ORDBMS features: OODBMS and RDBMS by allowing users to take
1) object view 2) object inheritance and 3) advantage of OODB'MSs great productivity and
object integration to solve data complexity complex data type without losing their existing
problems. Course content and students’ learning investment in relational data (Connolly & Begg,
outcomes are discussed. Many of the ORDBMS 2006). In fact, an ORDBMS engine supports both
features appear in Oracle. Thus, the author relational and object-relational features in an
utilizes Oracle as a tool to demonstrate how to integrated fashion (Frank, 1995). The underlying
overcome some weaknesses of relational DBMS. ORDB data model is relational because object
The ORDBMS script in the case study has been data is stored in tables or columns. ORDB
tested in the Oracle 9i, 10g, and 11g SQLPlus designers can work with familiar tabular
environment. The solution to the presented case structures and data definition languages (DDLs)
can be utilized in the classroom demonstration while assimilating new object-oriented features
and can also be generalized the homework (Krishnamurthy et al., 1999).It is essentially a
assignments and projects of advanced database relational data model with object-oriented
courses. extensions. In response to the evolutional
change of ORDB technology, SQL:1999 started
2. ORDB TECHNOLOGY supporting object-relational data modeling
features in database management
The object-relational database technology standardization and SQL:2003 continues this
occurrence can be traced back to the middle of evolution. Currently, all the major database
1990s after emergence of object-oriented vendors have upgraded their relational database
database (OODB). In their book “Object- products to object-relational database
relational DBMSs: the Next Great Wave”, management systems to reflect the new SQL
Stonebraker and Moore (1996) define their four- standards (Hoffer et al., 2009) and use by
quadrant view (two by two matrix) of the data industrial practitioners.
processing world: relational database, object-
relational database, data file processing, and Although each of the object-relational DBMS
object-oriented database. Their purpose is to vendors has implemented OO principles:
indicate the kinds of problems each of four- encapsulation and inheritance in their own way,
quadrants solves. As will be seen, "one size does all of them share the combination of the OO
not fit all"; i.e. there is no DBMS that solves all principles and follow SQL standardization,
the applications. They suggest that there is a incorporate object-oriented paradigms. All the
natural choice of data manager for each of the ORDBMSs have the ability to store object data
four database applications. They conclude why and methods in databases. Many of the
the problems addressed by object-relational SQL:2003 standard ORDBMS features appear in
DBMSs are expected to become increasingly Oracle. These features are listed as follows.
important over the next decade. As such, it is
"the next wave". Object Types: User-defined data types (UDT)
or abstract types (ADT) are referred to as object
Theoretically, as Stonebraker and Moore (1996) types.
predict in their four-quadrant view of the
database world, ORDBMS has been the most Functions/Methods: For each object type, the
appropriate DBMS that processes complex data user can define the methods for data access.
and complex queries. The object-oriented Methods define the behavior of data.
database management systems have made

©2011 EDSIG (Education Special Interest Group of the AITP) Page 81

www.aitp-edsig.org /www.isedj.org
Information Systems Education Journal (ISEDJ) 9 (4)
September 2011

Varray: The varray is a collection type that maintenance cost, reduce the defect rate, and
allows the user to embed homogenous data into raise the applications’ reliability. If multiple
an array to form an object in a pre-defined array database applications use the same set of
data type. database objects in ORDBMS, a de facto
standard for the database objects is created,
Nested table: A nested table is a collection type and these objects can be extended, reused and
that can be stored within another table. With a integrated in the ORDB.
nested table, a collection of multiple columns
from one table can be placed into a single 3. CASE STUDY
column in another table.
3.1 Case Scenario
Inheritance: With Object type inheritance,
users can build subtypes in hierarchies of Pacific Bike Traders assembles and sells bikes to
database types in ORDBs. customers. The company currently accepts
customer orders online and wants to be able to
Object View: Object view allows users to track orders and bike inventory. The existing
develop object structures in existing relational database system cannot handle the current
tables. It allows data to be accessed or viewed transaction volume generated by employees
in an object-oriented way even if the data are processing incoming sales orders. When a
really stored in a traditional relational format. customer orders a bike, the system must
confirm that the ordered item is in stock. The
There is some research that has been done in system must update the available quantity on
ORDBMS technology as ORDBMSs have become hand to reflect that the bike has been sold.
commonplace in recent years. He and Darmont When Pacific Bike Traders receives new
(2005) propose the Dynamic Evaluation shipments, a receiving clerk must update the
Framework (DEF) that simulates access pattern inventory to show the new quantity on hand.
changes using configurable styles of change. The system must produce invoices and reports
Pardede, Rahayu, & Taniar (2006) propose an showing inventory levels.
innovative methodology to store XML data into
new ORDB data structures, such as user-defined 3.2. Business Rules
type, row type and collection type. The
methodology has preserved the conceptual The following business rules are developed for
relationship structure in the XML data, including the new database system:
aggregation, composition and association. Wok
(2007) and Cho, et. al. (2007) present a One customer may originate many orders.
methodology for designing proper nesting One order must be originated from a customer.
structures of user-defined types in object-
relational database. The proposed schema trees One order must contain one or more bikes.
schema are transformed to Oracle 10g. Their One bike may be in many orders.
purpose is to develop an automatic ORDB design
tool. One employee may place many orders.
One order must be placed by an employee.
But very little research has been done in using
ORDBMS to overcome relational database One bike is composed with a front wheel, rear
weaknesses and solve some existing wheel, crank, and stem.
normalization problems. The significance of the One front wheel, rear wheel, crank, and stem
paper is to promote teaching ORDBMS features compose one bike.
for problem solving and object reuse and
integration among IS educators. The use of One employee must be either a full-time or part-
ORDBMSs to develop database applications can time.
enforce the reuse of varying user-defined object One full-time or part-time employee must be an
types, provide developers’ an integrated view of employee.
data and allow multiple database applications to
operate cooperatively. Ultimately, this can 3.2. ORDB Design
result in improved operational efficiency for the
IT department, increase programmers’ The Pacific Trader Object-Relational Database
productivity, lower development effort, decrease design is illustrated with the UML class diagram

©2011 EDSIG (Education Special Interest Group of the AITP) Page 82

www.aitp-edsig.org /www.isedj.org
Information Systems Education Journal (ISEDJ) 9 (4)
September 2011

in Appendix 1. Each of the classes is displayed Normalization Form (3NF) rule because there is
as a rectangle that includes three sections: the the transitive dependency in the customer table.
top section gives the class name; the middle Zip is a determinant of street, city and state.
section displays the attributes of the class; and Functional dependency analysis shows transitive
the last section displays methods that operate dependency:
on the data in the object. Associations between
classes are indicated with multiplicity Zip -> Street, City, State (transitive dependency)
(“min..max.” notation). Inheritance is indicated
with an empty triangle. Aggregation is marked There are three solutions to this transitive
with an empty diamond, whereas composition is dependency problem. Solution 1 keeps the
marked with a solid diamond. Aggregation customer table in the Second Normalization
models a whole-part relationship where Form (2NF) though it is not an ideal normal form
individual items become elements in a new for a relational database.
class. In Appendix 1, a sales order is made of
line items (bikes). Aggregation is indicated by a Solution 2 is to create a new customer address
small empty diamond next to the SalesOrder table by splitting the address from the original
class. The dotted line links to the associative customer table (3NF). This solution implies more
class generated from the many-to-many joins of records in the Customer table and Zip
relationship. table.

Based on the Pacific Trader’s Object-Relational Customer Table

Cu_id First Last Zip
Database Design in Appendix 1, ORDB features
1 John Smith 90201
are implemented with Oracle for the case in the
2 Mary Fox 92821
following sections. The implementation shows
how the UML class diagram maps and supports Zip Table
major ORDB features. For the sake of simplicity, Zip Street City State
it is assumed that referential integrity 96123 12 Pine Bell CA
constraints will be added later. 25678 6 Circle Brea VA

4. ORDBMS FOR NORMALIZATION Solution 3 is to store all the customer address

information in one column. This solution creates
Normalization is a logical data modeling difficulty in data retrieval. For example, it is
technique for the development of a well impossible to retrieve or sort customer records
structured relational database. The process is by city, state or zip code.
decomposing tables with anomalies to produce
smaller tables. Traditional normalization Customer table
processes are normalizing tables in non-1NF Cu_id First Last Address
form and multi-value attributes to at least 3NF; 1 John Smith 12 Pine, Bell, CA 90201
and removing transitive dependency. Such 2 Mary Fox 6 Circle, Brea, CA 92821
processes can be eliminated if ORDB technology
is used. None of the above three solutions is considered
ideal in terms of efficient database design and
4.1. Object Type & Transitive Dependency operations. The first solution is not satisfactory
since 2NF is not ideal for relational database
The address attribute is usually split into four design. The second solution implies that more
columns such as street, city, state and zip code joins might occur in the query process, since the
in order to store address dada in a customer zip table has been added to the database. The
table since it is a composite attribute in a third solution creates difficulty in data retrieval.
traditional database. For example, it is impossible to retrieve or sort
customer records by city, state, or zip code.
Customer table
Cu_id First Last Street City State Zip With ORDBMS technology, the attribute address
1 John Smih 12 Pine Bell CA 90201 can be defined as a user-defined abstract data
2 Mary Fox 6 Circle Brea CA 92821 type with a number of attributes using s the
same internal format. User-defined types (UDT)
The above Customer table is in Second or abstract data types (ADT) are referred to as
Normalization Form (2NF) and violates the Third object types. Object types are used to define

©2011 EDSIG (Education Special Interest Group of the AITP) Page 83

www.aitp-edsig.org /www.isedj.org
Information Systems Education Journal (ISEDJ) 9 (4)
September 2011

either object columns or object tables. The CUSTNAME.L_NAME CUSTADDRESS.CITY CU

following UML Customer class illustrates the
John Smith Bell CA
address object column.

SELECT * from Customer2;

Customer CUSTNAME(F_NAME
CUST_I CUSTADDRESS(STREET
-<PK> cust_id : Integer , L_NAME,
-name : Object D , CITY, STATE, ZIP)
INITIALS)
-address : Object
-<multivalued>phone : Object NAME_TY('John', ADDRESS_TY(’12 Pine’,
1
+getFullName() 'Smith') 'Bell', 'CA', 90201)

Object types need to be defined before the 4.2 Varray and Multi-value Attributes
customer table. The following SQL statements
define the object types: address_ty and In a relational model, multi-valued attributes
name_ty. are not allowed in the first normalization form.
The traditional solution to the problem is that
CREATE OR REPLACE TYPE address_ty AS each multiple-valued attribute is handled by
OBJECT forming a new table in a relational database. If a
(street NVARCHAR2(30), table has five multi-valued attributes, that table
city VARCHAR2(25), would have to be split into six tables. The Oracle
state CHAR(2), ORDBMS allows users to create the varying
zip NUMBER(10)); length array (VARRAY) data type as a new data
storage method for multi-valued attributes. The
CREATE OR REPLACE TYPE name_ty AS OBJECT following statement defines a varray type of
( three VARCHAR2 string named varray_phone_ty
f_name VARCHAR2(25), to represent a list of phone numbers.
l_name VARCHAR2(25));
VARRAY is a collection type in ORDBMSs. A
Mapping the above customer class, the following VARRAY consists of a set of objects that have
statement is used to create the Customer table the same predefined data type in an array. In a
with the CustName and CustAddress object relational model, multi-valued attributes are not
columns using name_ty and address_ty. The allowed in the first normalization form. The
column phone is to be added to the table later. solution to the problem is that each multiple-
valued attribute is handled by forming a new
CREATE TABLE Customer2( table. If a table has five multi-valued attributes,
Cust_ID Number(5), that table would have to be split into six tables
CustName name_ty, after the First Form of normalization. To retrieve
CustAddress address_ty); the data back from that original table, the
student would have to do five joins across these
Object type constructors are used to insert six tables. ORDBMs allow multi-valued attributes
object data into the table. The following INSERT to be represented in a database. ORDBMSs allow
statement uses constructors name_ty() and users to create the varying length array
address_ty() to add data into the two object (VARRAY) data type can be used as a new data
columns. storage method for multi-valued attributes. The
following statement defines a VARRAY type of
INSERT INTO Customer VALUES (1, three VARCHAR2 strings named
name_ty ('John', 'Smith',), varray_phone_ty to represent a list of three
address_ty ('12 Road', 'Bell', 'CA', 90201)); phone numbers in the Customer2 table.

The following statements retrieve the data from CREATE TYPE varray_phone_ty AS VARRAY(3)
the Customer2 table. OF VARCHAR2(14);

SELECT c.custName.l_name, c.custAddress.City, ALTER TABLE Customer ADD (phones

c.custAddress.state varray_phone_ty);
FROM Customer2 c;

©2011 EDSIG (Education Special Interest Group of the AITP) Page 84

www.aitp-edsig.org /www.isedj.org
Information Systems Education Journal (ISEDJ) 9 (4)
September 2011

UPDATE customer The following statement creates nested table

SET phones = types: wheel_type, crank_type and stem_type:
(varray_phone_ty('(800)555-1211',
'(800)555-1212','(800)555-1213')) CREATE TYPE nested_table_wheel_type AS
WHERE cust_id = 1; TABLE OF wheel_type;

INSERT INTO customer(phones) values CREATE TYPE nested_table_crank_type AS

(varray_phone_ty('(800)555- TABLE OF crank_type;
1211','(800)555-1212','(800)555-1213'));
CREATE TYPE nested_table_stem_type AS
The above example shows that using the varying TABLE OF stem_type;
length array (VARRAY) data type not only can
solve multi-value attribute problem for the The following example creates the table named
customer table, but also can speed up the query Bike with that contains four nested tables:
process on customer data.
CREATE TABLE bike (
4.3 Nested Table and Non-1NF serial_no INTEGER PRIMARY KEY,
model_type VARCHAR2(20),
A nested table is a table that can be stored front_wheel
within another table. With a nested table, a nested_table_wheel_type,
collection of multiple columns from one table can rear_wheel
be placed into a single column in another table. nested_table_wheel_type,
Nested tables allow user to embed multi-valued crank
attributes into a table, thus forming an object. nested_table_crank_type,
stem
nested_table_stem_type
)
Bike
-<PK>serial_no : Integer
NESTED TABLE
-model_type : String front_wheel
-qty_on_hand : Integer
-list_price : Decimal STORE AS
+updateInventory() front_wheel,
+getBike()
NESTED TABLE
1 rear_wheel
STORE AS
rear_wheel,
1 1
2 NESTED TABLE
Wheel Crank Stem crank
-SKU : Integer -SKU : Integer -SKU# : Integer STORE AS
-rim : String
-spoke : String
-size : String
-weight : String
-size : String
-weight : String nested_crank,
-tire : String +getCrank() +getStem() NESTED TABLE
+getWheel()
stem
STORE AS
CREATE TYPE wheel_type AS OBJECT( nested_stem;
SKU VARCHAR2(15),
rim VARCHAR2(30), INSERT INTO bike VALUES (1000, 'K2 2.0 Road',
spoke VARCHAR2(30), nested_table_wheel_type( wheel_type('w7023',
tire VARCHAR2(30)); '4R500', '32 spokes', '700x26c' )),
nested_table_wheel_type(
CREATE TYPE crank_type AS OBJECT wheel_type('w7023', '4R500', '32 spokes',
(SKU VARCHAR2(15), '700x26c' )),
crank_size VARCHAR2(15), nested_table_crank_type(
crank_weight VARCHAR2(15) ); crank_type('c7023', '30X42X52', '4 pounds')),
nested_table_stem_type(
CREATE TYPE stem_type AS OBJECT( stem_type('s7023', 'M5254', '2 pounds')));
SKU VARCHAR2(15),
stem_size VARCHAR2(15), Finally the previous statement inserts a row into
stem_weight VARCHAR2(15)); the Bike table with nested tables using the three

©2011 EDSIG (Education Special Interest Group of the AITP) Page 85

www.aitp-edsig.org /www.isedj.org
Information Systems Education Journal (ISEDJ) 9 (4)
September 2011

defined constructors: wheel_type, crank_type INSERT INTO SalesOrder VALUES

and stem_type. (100,'5-Sep-05', 1, '1000');
INSERT INTO salesOrder VALUES
The above example shows that using the (101, '1-Sep-05', 1, '1000');
NESTED TABLE can implement the composition
association, store multiple parts and also speed The following statements show how to create an
up the data retrieval speed for the Bike table. object view on the top of the SalesOrder
The following statement shows the nested tables relational table:
in the table Bike.
CREATE TYPE SalesOrder_type AS OBJECT(
SELECT * from bike; sales_ord_id NUMBER(10),
ord_date DATE,
cust_id NUMBER(10),
emp_id NUMBER(10));

CREATE VIEW customer_order_view OF

SalesOrder_type WITH OBJECT IDENTIFIER
(sales_ord_id)
AS SELECT o.ord_id, o.ord_date, o.cust_id,
o.emp_id
FROM salesOrder o
WHERE o.cust_id = 1;

The following SQL statement generates the view

5. ORDBMS FOR OBJECT INTEGRATION
output:
The beauty of ORDBMSs is reusability and
SELECT * FROM customer_order_view;
sharing. Reusability mainly comes from storing
data and methods together in object types and
performing their functionality on the ORDBMS
server, rather than have the methods coded
separately in each application. Sharing comes
from using user-defined standard data types to
make the database structure more standardized The object view is a bridge that can be used to
(Breg & Connolly. 2010) create object-oriented applications without
modifying existing relational database schemas.
5.1. Object Views on a Relational Table By calling object views, relational data can be
retrieved, updated, inserted, and deleted as if
Object views are virtual object tables, which such data were stored as objects. The following
allow database developers to add OOP structures statement can retrieve Analysts as object data
on top of their existing relational tables and from the relational SalesOrder table. Using
enable them to develop OOP features with object views to group logically-related data can
existing relational data. The object view is a lead to better database performance.
bridge between the relational database and OOP.
Object view creates a layer on top of the 5.2 Inheritance for Object Reuse
relational database so that the database can be
viewed in terms of objects (Loney & Koch, The main advantages of extending the relational
2002). This enables you to develop OOP features data model come from reuse and sharing. If
with existing relational data. The following multiple applications use the same set of
statements show how to create the SalesOrder database objects, then you have created a de
table: facto standard for the database objects, and
these objects can be extended (Price, 2002).
CREATE TABLE SalesOrder ( ORDBMSs allow users to define hierarchies of
ord_id NUMBER(10), data types. With this feature, users can build
ord_date DATE, subtypes in hierarchies of database types. If
cust_id NUMBER(10), users create standard data types to use for all
emp_id NUMBER(10)); employees, then all of the employees in the
database will use the same internal format.

©2011 EDSIG (Education Special Interest Group of the AITP) Page 86

www.aitp-edsig.org /www.isedj.org
Information Systems Education Journal (ISEDJ) 9 (4)
September 2011

Users might want to define a full time employee CREATE TYPE full_time_ty UNDER employee_ty (
object type and have that type inherit existing Salary NUMBER(8,2));
attributes from employee_ty. The full_time_ty
type can extend employee_ty with attributes to CREATE TABLE FullTimeEmp of full_time_ty;
store the full time employee’s salary. The
part_time_ty type can extend employee_ty with The preceding statement creates full_time_typ
attributes to store the part-time employee’s as a subtype of employee_typ. As a subtype of
hourly rates and wages. Inheritance allows for employee_ty, full_time_ty inherits all the
the reuse of the employee_ty object data type. attributes declared in employee_ty and any
The details are illustrated in the following class methods declared in employee_ty. The
diagram: statement that defines full_time_ty specializes
employee_ty by adding a new attribute “salary”.
New attributes declared in a subtype must have
Employee
-<PK> emp_id : Integer
names that are different from the names of any
-name : Object attributes or methods declared in any of its
-address : Object
-phone : Object
supertypes, higher up in its type hierarchy. The
+getEmployee() following example inserts row into the
FullTimeEmp table. Notice that the additional
salary attribute is supplied

INSERT INTO FullTimeEmp VALUES

FullTime PartTime
(1000, 123456789, name_ty('Jim', 'Fox', 'K'),
-rate : Decimal
-salary : Decimal
-hours : Integer '12-MAY-1960',
+updateSalary()
+updateRate() varray_phone_ty('(626)123-5678', '(323)343-
2983', '(626)789-1234'),
Object type inheritance is one of new features of Address_ty ('3 Lost Spring Way', 'Orlando', 'FL',
Oracle 9i. For employee_ty to be inherited from, 32145), 45000.00);
it must be defined using the NOT FINAL clause
because the default is FINAL, meaning that SELECT * FROM FullTimeEmp;
object type cannot be inherited. Oracle 9i can
also mark an object type as NOT
INSTANTIABLE; this prevents objects of that
type derived. Users can mark an object type as
NOT INSTANTIABLE when they use the type only
as part of another type or as a super_type with
NOT FINAL. The following example marks
address type as NOT INSTANTIABLE:

CREATE TYPE employee_ty AS OBJECT (

A supertype can have multiple child subtypes
emp_id NUMBER,
called siblings, and these can also have
SSN NUMBER,
subtypes. The following statement creates
name name_ty,
another subtype part_time_ty under
dob DATE,
Employee_ty.
phone varray_phone_ty,
address address_ty
CREATE OR REPLACE TYPE part_time_ty UNDER
) NOT FINAL NOT INSTANTIABLE;
employee_ty (
rate Number(7,2),
To define a new subtype full_time_ty inheriting
hours Number(3))NOT FINAL;
attributes and methods from existing types,
users need to use the UNDER clause. Users can
CREATE TABLE PartTimeEmp of part_time_ty;
then use full_time_ty to define column objects
or table objects. For example, the following
A subtype can be defined under another
statement creates an object table named
subtype. Again, the new subtype inherits all the
FullTimeEmp.
87attributes and methods that its parent type
has, both declared and inherited. For example,
the following statement defines a new subtype

©2011 EDSIG (Education Special Interest Group of the AITP) Page 87

www.aitp-edsig.org /www.isedj.org
Information Systems Education Journal (ISEDJ) 9 (4)
September 2011

student_part_time _ty under part_time_ty. The Object-types possess all the properties of a
new subtype inherits all the attributes and class, data abstraction, encapsulation,
methods of student_part_time _ty and adds two inheritance and polymorphism. These traits of
attributes. object-types are embedded in the relational
nature of the database; data model, security,
CREATE TYPE student_part_time_ty UNDER concurrency, normalization. In more precise
part_time_ty words, the underlying ORDB data model is
(school VARCHAR2(20), relational because object data is stored in tables
year VARCHAR2(10)); or columns.

5.3 Object Integration with Interface 6. LEARNING OUTCOMES

ORDBMS combines attributes and methods The provided ORDB script guides students with
together in the structure of object type. The hands-on learning experience in the classroom.
object type interface includes both attributes Once they have understood they can use the
and its methods. The public interface declares script as templates to do their homework
the data structure and the method header shows assignments and projects. ORDB, implement it
how to access the data. This public interface with Oracle 9i/10g, and create ORDB
serves as an interface to applications. The applications using various tools. As a result, the
private implementation fully defines the following learning outcomes are demonstrated at
specified methods. the end of the class. Students are able to:

Public Interface 1. Map UML class diagrams to ORDB databases

Specification: 2. Use Object Types to remove transitive
Attribute declarations dependency
Method specifications 3. Use VARRAY types for multi-value attribute
4. Use NESTED TABLE types to Solve non-1NF
Private Implementation problems
Body: 5. Implement inheritance with sub-object types
Method implementations 6. Create object views in the existing relational
databases
The following statement displays the public
interface of the object type name_type. The ORDB technology helps students to better
output of the name_type public interface shows understand object-oriented principles such as
attributes and method headers as follows: encapsulation, inheritance, and reusability.
During the learning process, they have reviewed
DESC name_ty; the object-oriented paradigm they learned from
their previous programming courses and are
able to tie it to ORDBMS and object-oriented
system design.

With a grasp of ORDB technology, students are

able to make their database design more
METHOD structured and consistent. With object reuse and
MEMBER FUNCTION FULL_NAME standard adherence, students are able to create
RETURNS VARCHAR2 a de facto standard for database objects and
multiple database applications. The motivation
Although the user-defined methods are defined to learn in class is high because students have
with object data within the object type, they can realized that object-relational technology is
be shared and reused in multiple database incorporated in most commercial DBMS.
application programs. This can result in Learning it will help their career development in
improved operational efficiency for the IT the future competitive job market.
department, as well, by improving
communication and cooperation between 7. REFERRNCES
applications. An object-relational database
schema consists of a number of related tables Begg, C., & Connolly, T. (2010). Database
that forms connected user-defined object-types. systems: A practical approach to design,

©2011 EDSIG (Education Special Interest Group of the AITP) Page 88

www.aitp-edsig.org /www.isedj.org
Information Systems Education Journal (ISEDJ) 9 (4)
September 2011

implementation, and management, 5th Ed. mainstream, Proceedings of the ACM

Addison Wesley. SIGMOD International Conference on
Management of Data and Symposium on
Cho, W., Hong, K. & Loh, W. (2007). Estimating Principles of Database Systems,
nested selectivity in object-oriented and Philadelphia, PA
object-relational databases Information and
Software Technology, (49)7, 806-816 Loney, K. & Koch, G. (2002) Oracle 9i: The
complete reference, Oracle Press/McGraw-
Connolly, T. and Begg, C. (2006). Database Hill/Osborne.
systems: A practical approach to design,
implementation, and management, 4th Ed. Mok, W. Y. (2007) Designing nesting structures
Addison Wesley. of user-defined types in object-relational
databases, Information and Software
Elmasri, R. & Navathe, S. (2011). Fundamentals Technology, 49, 1017–1029.
of Database Systems, 6th Edition, Addison
Wesley. Pardede, E., Rahayu, J. Wenny, T. & Taniar, D.,
2006, Object-relational complex structures
Fortier, P. (1999). SQL3: Implementing the for XML, Information & Software Technology,
Object-Relational Database, Osborne 48(6), 370-384.
McGraw-Hill,
Philippi, S. 2005, Model driven generation and
Frank, M. (1995). Object-relational Hybrids, testing of object-relational mappings,
DBMS, 8/8, 46-56. Journal of Systems and Software, 77:2, 193-
207.
Garcia-Molina, H., Ullman, J. & Widom, J. 2003.
Database Systems: The Complete Book, Price, J. 2002. Oracle9i, JDBC Programming,
Prentice Hall, Upper Saddle River. Oracle Press/McGraw-Hill/Osborne

He, Z., & Jérôme, D. (2005). Evaluating the Rahayu, J. W., Taniar, D. And Pardede, E.
Dynamic Behavior of Database Applications, (2005) Object-Oriented Oracle, IRM Press
Journal of Database Management; 16:2, 21-
45. Silberschatz, A., Korth, H. and Sudarshan, S.
2009, Database System Concepts, Six
Hoffer, J., Prescott, M., & Topi, H., 2009 Modern Edition, McGraw-Hill
Database Management, 9th Edition, Pearson
Prentice Hall. Stonebraker M. and Moore, D. 1996. Object-
relational DBMSs: the Next Great Wave. San
Krishnamurthy, Banerjee and Nori, 1999. Francisco, CA: Morgan Kaufmann Publishers,
Bringing object-relational technology to the Inc.

©2011 EDSIG (Education Special Interest Group of the AITP) Page 89

www.aitp-edsig.org /www.isedj.org
Information Systems Education Journal (ISEDJ) 9 (4)
September 2011

Appendix 1 Pacific Trader’s Object-Relational Database Design

©2011 EDSIG (Education Special Interest Group of the AITP) Page 90

www.aitp-edsig.org /www.isedj.org