COMP67321

Introduction to Object-oriented Databases

Goran Nenadic School of Computer Science University of Manchester
1

Aims
Understand the need for object oriented DBs Learn basic concepts and principles Learn the basics of ODL and OQL Understand advantages and disadvantages of object databases

Plan
Overview of object-oriented methodology Object data model (ODMG) Object definition language (ODL) Object query language (OQL) Object and object-relational DBs

COMP67321, 2010/2011 School of Computer Science

Problems with relational DBs

Focused on traditional business applications Limited data types and operators
limited data abstraction

Definition of additional specific types and operators is not straightforward

need to be embedded in application programs

Insulation between data and operations

continued

Problems with relational DBs

Integration with OO languages
storage of complex objects from a Java program is not straightforward

Need for
abstract data types encapsulation of operations inheritance better integration with programming languages

OODBs introduction
Integration of two technologies: OO programming and DBs Main ideas
raise the level of abstraction provide direct correspondence between real-world and database objects objects do not lose their integrity and identity and can easily be identified and operated upon

COMP67321, 2010/2011 School of Computer Science

OODBs objects
An object belongs to a specific class Have structure, state (value), behavior (operations), identity, name(s) Object structure of arbitrary complexity
in order to contain all of the necessary information

Contrast to traditional database systems

information about a complex object is often scattered over many relations or records, leading to loss of direct correspondence between a real-world object and its database representation
7

OODBs operations
Applied to objects of a particular type
includes non-standard application-specific operations

Operations can be invoked by passing a message to an object

includes the operation name and the parameters

Operator polymorphism/overloading
ability to be applied to different types of objects operation name may refer to several distinct implementations, depending on the type of objects it is applied to

Brief overview of OO concepts

Classes and objects Object identity Encapsulation and information hiding Inheritance Polymorphism
(see additional materials on the Web)

COMP67321, 2010/2011 School of Computer Science

OODB standards, models and languages

10

Why OO standards?
One of the reasons for success of the relational database paradigm was the SQL standard
portability, interoperability

Object Data Management Group proposed a standard ODMG that includes

Object Model (OM) Object Definition Language (ODL) Object Query language (OQL) programming language bindings

11

Note: ODMG
The ODMG is a non-profit consortium of database vendors and interested parties who collaborated to define data storage portability standards for object-oriented applications. Since its inception in 1991, the ODMG has continued to grow and today includes database vendors, tool vendors, consulting firms, and corporate end users. Member companies include Advanced Language Technologies, Andersen Consulting, Ardent Software, Baan, CERN, Computer Associates, Ericsson, GemStone Systems, Hitachi, IBEX Computing SA, JavaSoftSun Microsystems, Lockheed Martin, Microsoft, NEC Corporation, Object Design, The Object People, Objectivity, POET Software, Sybase, Telenor R&D, Versant Object Technology, and Watershed Technologies. The ODMG group completed its work on object data management standards in 2001 and was disbanded. The final release of the ODMG standard (ODMG 3.0, including the overall data model) can be found at http://www.odmg.org 12

COMP67321, 2010/2011 School of Computer Science

ODMG object model

Provides a standard data model for object databases Supports object definition via ODL Supports object querying via OQL Supports a variety of data types and type constructors

13

continued

ODMG object model

basic concepts
objects
collections atomic objects (user-defined)

literals interface [see additional materials] class extent, key, factory objects
14

ODMG objects
Object has four characteristics
identifier (OID): unique system-wide identifier name: unique within a particular database and/or program; [optional] lifetime: persistent vs. transient structure: specifies how object is constructed by the type constructor and whether it is collection object atomic object

15

COMP67321, 2010/2011 School of Computer Science

ODMG built-in collections

Set unordered collection that does not allow duplicates Bag unordered collection that allows duplicates List ordered collection that allows duplicates Array one-dimensional array of dynamically varying length, with direct access Dictionary unordered sequence of key-value pairs with no duplicate keys
Notes: 1) all objects in a collection must be of the same type 2) OM defines built-in interfaces for each collection (see later) 16

ODMG atomic objects

Any object that is not a collection
typically structured complex objects, such as tuples (struct)

Contains
properties: attributes and relationships operations

Attribute values can be literals or OIDs Note

atomic objects are defined via keyword class in ODL (see later)

17

ODMG literals
Literal has a current value but not an identifier
embedded in objects

Three types of literals

atomic: basic (predefined) data type values (e.g., short, float, boolean, char) structured: values that are constructed by type constructors (e.g., date, time, struct variables) collection: a collection (e.g., set, array) of values or objects

18

COMP67321, 2010/2011 School of Computer Science

ODMG class
Class is a specification of abstract behaviour and abstract state of an object
classes are instantiable

Inheritance through classes allows both state and behavior inheritance, but (strictly) among classes Two concepts related to classes: extent and key

19

continued

ODMG class
Extent of a class
contains all persistent objects of that class similar to creating an object of type Set<class_name> and making it persistent

Keys
one or more properties whose values are unique for objects in an extent e.g. national insurance number

20

ODMG factory object

Used to generate or create individual objects Basically, provides constructor operations for new objects There is an interface called ObjectFactory
interface ObjectFactory { Object new (); };

new( ) returns new objects with an object_id

21

COMP67321, 2010/2011 School of Computer Science

ODMG database
ODMG Object Model supports the concept of databases as storage areas for persistent objects of a given set of types (defined by OM) A database has a schema that contains a set of type definitions (see ODL) ODMG defines interfaces for Database and DatabaseFactory objects

22

continued

ODMG database
Each OO database has
its database name bind operation (assigns individual unique names to persistent objects)
named objects are entry points to the database

unbind (removes bound names) lookup (retrieves an object with the specified name)

23

24

COMP67321, 2010/2011 School of Computer Science

Object Definition Language (ODL)

25

ODL
Supports semantics constructs of ODMG Independent of any programming language
but can be easily mapped to a OO language using the specific language bindings

Used to create object specifications (classes and interfaces) Not used for database manipulation

26

ODL class definition

class Name1 [ extends Name2 ] ( { attribute struct s_name {.} name; attribute string; . relationship class_name r_name inverse name; . void a_method(); }
27

extent name_of_entext key name_of_key )

COMP67321, 2010/2011 School of Computer Science

Class definition example 1

class Person ( extent persons key ssn ) { attribute struct attribute string attribute date attribute enum attribute struct short age(); }

Pname {string fname, string lname} name; ssn; birthdate; Gender{M, F} sex; Address {short no, string street, string city, string post_code} address;

28

Class definition example 2

Class FacultyMember extends Person (extent faculty) { attribute string rank; attribute float salary; attribute string phone; relationship Dept works_in inverse Dept::has_facultyMember; relationship set<GradStu> advises inverse GradStu::advisor; void give_raise (in float raise); void promote (in string new_rank); };
29

Class definition inheritance

Inheritance is done using extends
individual objects inherit both the properties (attributes and relationships) and operations

Multiple class inheritance is not supported! (but multiple behaviour inheritance is)

30

COMP67321, 2010/2011 School of Computer Science

10

Class definition relationships

Relationships: a property that specifies that two objects are related Only binary relationships are explicitly represented in the object model Inverse relationship used to specify a single conceptual relationship (possibly in both ways)
relationship Dept works_in inverse Dept::has_facultyMember; FacultyMember works_in Dept Dept has_facultyMember FacultyMember

31

continued

Class definition relationships

Similar construction for Dept class
Class Dept ( extent departments key dname ) { attribute string dname; attribute FacultyMember head; . relationship set<FacultyMember> has_facultyMember inverse FacultyMember::works_in; };

inverses are used to maintain referential integrity

32

continued

Class definition relationships

If a relationship is to be represented only in one direction use attributes
attribute FacultyMember chair;

Syntax for specifying collections: collection<class>

Class FacultyMember extends Person { relationship set<GradStu> advises inverse GradStu::advisor; };

A FacultyMember advises a set of GradStus

33

COMP67321, 2010/2011 School of Computer Science

11

Class definition operations

Operation signatures
name, argument types and return value

Operation names are unique within the class/interface Can specify exceptions
raising exceptions during operation execution
void register_student(in string ssn) raises(student_no_valid, section_full);

34

Note: language bindings

Specify how ODL constructs are mapped to constructs in a specific OO language
Smalltalk and C++ bindings defined by ODMG (see Ch 21.4 for C++) ODMG Java binding has been superseded by Java Data Objects (JDO)

35

36

COMP67321, 2010/2011 School of Computer Science

12

Object Query Language (OQL)

37

OQL
ODMGs query language Syntax is similar to SQL with additional features for objects Works closely with programming languages such as C++
embedded OQL statements return objects that are compatible with the type system of the host language

38

Basic syntax
SELECT d.name FROM d in departments WHERE d.college = Engineering;

An entry point to the database (e.g. named persistent object) is needed for each query Typically, an extent name (e.g., departments) may serve as an entry point
Recall: an extent is a persistent set of all persistent objects from a class
39

COMP67321, 2010/2011 School of Computer Science

13

Basic syntax: iterators

Iterator variables (d) are defined whenever a collection is referenced in an OQL query
in the previous example, d ranges over each object in the collection (departments)

Syntactical options for specifying an iterator:

d in departments departments d departments as d

40

Basic syntax: data type of result

bag: selectfromwhere
previous example: bag<string>

set: select distinct fromwhere list: select fromwhereorder by

41

continued

Basic syntax: data type of result

However, a query does not have to follow the selectfromwhere format The data type of a query result can be any type defined in the ODMG model E.g., a persistent name on its own can serve as a query whose result is a reference to the persistent object (e.g., departments) whose type is set<Departments>

42

COMP67321, 2010/2011 School of Computer Science

14

OQL syntax summary

SELECT [DISTINCT] <expression> FROM <fromList> [WHERE <expression>] [GROUP BY <attribute1:expression1, attribute2:expression2,>] [HAVING <predicate>] [ORDER BY <expression>]

where: <fromList> ::= <variableName> IN <expression>| <variableName> IN <expression>,<fromList>| <expression> AS <variableName>| <expression> AS <variableName>,<fromList>
43

Basic syntax: path expressions

A path expression is used to specify a path to attributes and objects in an entry point A path expression starts at a persistent object name (or its iterator variable) The name will be followed by zero or more dot connected relationship or attribute names informatics_dept.chair
(assuming that there is a persistent object named informatics_dept)

44

continued

Basic syntax: path expressions

examples
select f.rank from f in informatics_dept.has_facultyMember select distinct f.rank from f in informatics_dept.has_facultyMember

FacultyMember

set<FacultyMember>

45

COMP67321, 2010/2011 School of Computer Science

15

Example class (Department)

Class Deptartment ( extent departments key dname ) { attribute string dname; attribute FacultyMember head; . relationship set<FacultyMember> has_facultyMember inverse FacultyMember::works_in; relationship set<Students> has_major inverse Student::majors_in; };

46

Example class (Student)

47

Example
select struct( last_name: s.name.lname, first_name: s.name.fname, gpa: s.gpa), from s in informatics_dept.has_majors where s.class = senior order by gpa desc, last-name asc, first_name asc; Retrieve last and first names plus GPA (average) of all senior (i.e. final year) students in the InformaticsDept; order the results by gpa, and then by names
48

COMP67321, 2010/2011 School of Computer Science

16

continued

Example
select struct( last_name: s.name.lname, first_name: s.name.fname, gpa: s.gpa), from s in students where s.majors_in.dname = Informatics and s.class = senior order by gpa desc, last-name asc, first_name asc; Retrieve last and first names plus GPA (average) of all senior (i.e. final year) students in the InformaticsDept; order the results by gpa, and then by names
49

Example for you

class Student (extent students key student_id) { attribute string name; attribute int student_id; attribute int year; relationship <Dept> studies_in inverse Dept::has_students; relationship <Staff> has_tutor inverse Staff::tutees; }

50

Example for you

class Staff (extent staff key staff_id) { attribute string name; attribute int staff_id; attribute string position; relationship set<Student> tutees inverse Student::has_tutor; }

51

COMP67321, 2010/2011 School of Computer Science

17

Example for you

1) Create a Dept class. 2) Retrieve names of all students in the InformaticsDept whose tutor is a senior lecturer. select s.name from s in informaticsDept.has_students where

52

Views as named objects

The define keyword in OQL is used to specify an identifier for a named query
name should be unique; if not, the results will replace an existing named query

Once a query definition is created, it will persist until deleted or redefined A view definition can include parameters

53

continued

Views as named objects

A view to include students in a department who have a minor (in that department):
define has_minor(dept_name) as select s from s in students where s.minor_in.dname = dept_name has_minor can now be used in queries: has_minor(Informatics) returns bag of students
54

COMP67321, 2010/2011 School of Computer Science

18

Aggregate operators
OQL supports a number of aggregate operators (that operate over a collection) that can be applied to query results min, max, count, sum, and avg count returns an integer; others return the same type as the collection type
count (s in has_minors(Informatics)); count (select f.name from f in faculty where f.sex = F);

55

continued

Aggregate operators
avg (select s.gpa from s in students where s.class = senior and s.majors_in.dname =Business);

56

Membership and quantification

(e in c) is true if e is in the collection c
Informatics in (select d.dname from d in departments) select s.name.fname, s.name.lname from s in students where COMP67321 in (select c.cname from c in s.completed_sections.section.of_course);
57

COMP67321, 2010/2011 School of Computer Science

19

continued

Membership and quantification

(for all e in c: b) is true if all e elements of collection c satisfy b (exists e in c: b) is true if at least one e in collection c satisfies b
for all g in (select s from s in grad_students where s.majors_in.dname = Informatics) : g.advisor in InformaticsDept.has_faculty;

58

Grouping operator
select deptname, avg_gpa : avg (select p.s.gpa from p in partition) from s in students group by deptname: s.majors_in.dname having count (partition) > 100;
expression attribute

Retrieve average GPA of majors in each department having >100 majors

59

Examples embedded query

select struct (name: struct( last_name: s.name.lname, first_name: s.name.fname), degrees: (select struct (deg: d.degree, yr: d.year, col: d.college) from d in s.degrees) from s in informatics_dept.chair.advises; Retrieve last and first names plus degree information for all students advised by the head of InformaticsDept
60

COMP67321, 2010/2011 School of Computer Science

20

Designing Object Databases

61

OODBs vs. RDBs: relations

OODB
relationships are handled by reference attributes that include OIDs of related objects single and collection of references are allowed references for binary relationships can be expressed in a single direction or both directions via inverse operator
if in both directions: redundancy, and hence inconsistency sometimes better to create a separate class to represent the relationship (cf. Grade, Student, Section classes)

62

continued

OODBs vs. RDBs: relations

RDB
relationships among tuples are specified by attributes with matching values (via foreign keys) foreign keys are single-valued M:N relationships must be presented via a separate relation (table)

63

COMP67321, 2010/2011 School of Computer Science

21

OODBs vs. RDBs: inheritance

Inheritance structures are built in OODB (via : and extends operators) RDB has no built-in support for inheritance relationships
object-relational databases

64

OODBs vs. RDBs: design

the specification of operations
OODB: operations specified during design (as part of class specification) RDB: may be delayed until implementation

65

Mapping EER to OODB schema

Mapping EER schemas into OODB schemas is relatively simple especially since OODB schemas provide support for inheritance relationships Once mapping has been completed, operations must be added to OODB schemas since EER schemas do not include an specification of operations

66

COMP67321, 2010/2011 School of Computer Science

22

Current status and conclusions

67

Current status
OODB market around $250M as opposed to the relational DB revenue of $20-50B OODBMS were originally thought of to replace RDBMS because of their better fit with objectoriented programming languages. However, high switching cost and the inclusion of objectoriented features in RDBMS to make them object-relational have made RDBMS successfully defend their dominance.

68

continued

Current status
Object databases are now established as a complement rather than a replacement for relational databases. OO ideas are being used in a large number of applications, without explicitly using the OODB platform to store data
OO tools for modelling and analysis, OO programming languages object relational DBMS

69

COMP67321, 2010/2011 School of Computer Science

23

continued

Current status
Database services

RDBMS RDBMS

Object-relational DBMS

File & Network File & Network Systems System

ODBMS OODBMS

Data modeling

70

Current status ORDBs

An object-relational database (ORDB) is a relational database that allows developers to integrate the database with their own custom data types and methods. Also, software running over traditional DBMSs to provide similar features The goal of ORDBMS technology is to allow developers to raise the level of abstraction at which they view the problem domain. Object-oriented features added to SQL-99
71

continued

Current status ORDBs

SQL-99 includes
some type construction (row, array, set, list) object identity (through reference types) encapsulation of operations inheritance new data types (BLOBs, CLOBs)

Many vendors provide specific type extensions for newer data type (e.g. cartridges in Oracle) Other vendors: Informix Universal Server, IBMs UDB, DB2/II
72

COMP67321, 2010/2011 School of Computer Science

24

Current status OODB products

Commerical
db4objects (db4o) GemStone (GemStone/S, Facets) InterSystems (Cach) Matisse (Matisse) ObjectDB (ObjectDB) Objectivity (Objectivity/DB) Progress (ObjectStore, PSE Pro) Versant (merged with Poet; VDS, FastObjects) W3apps (Jeevan)

Open source
db4o Ozone Perst Zope (ZODB)

73

Summary: OODB approach

User-defined data types and operations
this is part of the database definition direct correspondence between real-world and database objects abstract data types, inheritance, encapsulation

Integration and compatibility with OO languages

OODBs provide persistent and straightforward storage for program objects and data structures (not straightforward in RDBs) objects do not lose their integrity and identity
74

continued

Summary: OODB approach

main concepts/advantages (summary)
object identity support for complex object structures encapsulation of properties and operations type hierarchies and behaviour inheritance polymorphism programming language compatibility (storage of objects and ODL/OQL bindings)

75

COMP67321, 2010/2011 School of Computer Science

25

continued

Summary: OODB approach

Issues
OM is more complex than relational model
can represent more complex data structures than the simple relational representation, but more difficult to design representation of non-binary relationships not trivial

no early standards, so limited use, popularity, portability

low investment in non-standard technologies RDBs are traditionally much more used

OQL needs named entities

Object-relational databases as a compromise

76

Summary
Overview of OO methodology Object model and various constructs and built-in types of the ODMG model presented ODL and OQL languages were presented Current state of OODBs/ORDBs, advantages and disadvantages discussed

77

Reading for this lecture

Chapters 20 and 21 in the textbook Further information/optional reading: Object relational DBs: Chapter 22 http://www.odbms.org/, http://www.odmg.org/

78

COMP67321, 2010/2011 School of Computer Science

26