Object Databases

1
Object Databases

• ???
• OCIT
• pychang_at_ocit.edu.tw

2
Reference

• 1.Object-Relational DBMSstracking the next
  great wave by Michael. StoneBraker Paul.
  Brown,1999
• 2.Object Databases An ODMG Approach by Richard
  Cooper,1997
• 3.Fundation for Object/Relational Databases the
  third Manifesto, by C. J. Date, 1998

3
Chapter 1

• Developments in database technology

4
What is a database system?

• A DBMS is a coordinated package of software which
  permits large amount of structured data to be
  managed.

5
Facilities of a DBMS

• Application generality. The DBMS will be able to
  handle data from many different application areas
  equally easily.
• Efficient data access. The data will be stored in
  such a way that it can be accessed as quickly as
  possible.
• Security. There will be privacy mechanisms
  securing the data against unauthorized access and
  misuse.
• Data consistency Any changes made to the data
  will be organized in such a way that the database
  always remains consistent, i.e. no changes will
  be allowed to violate the organization imposed on
  the data by the database designer.

6
Facilities of a DBMS

• Resilience. Hardware or software failures while
  the database is running are never allowed to make
  the database inconsistent - at most a few of the
  most recent changes will be lost if the system
  crashes.
• Concurrency control. Many users can access the
  database at the same time without interfering
  with each others work.
• Multiple views. There will be a facility for
  providing apparently different or more focused
  representations to different kinds of user.
• Integrity. There may be the facility for imposing
  integrity constraints which limit the values
  which the database can take - thus maintaining
  the meaning of the information stored
• Distribution. It may be possible for the DBMS to
  handle data which is distributed across more than
  one computer joined in a local network or even
  across a number of sites joined in a wider
  network.

7
Facilities of a DBMS

• User interfaces. Appropriate user interfaces will
  be provided to allow databases to be designed and
  used by a wide variety of users. Among these will
  be
• Parametric interfaces. Eg. Automated teller
  machines (ATMs) provide the most prevalent
  example of this kind of interface.
• Ad hoc querying. The DBMS will provide a general
  purpose query language which gives users a fairly
  simple textual method for specifying the database
  structure updates to data and retrievals of data
  from the database.
• Graphical interfaces. In order to simplify the
  interaction, there are a variety of emerging
  interface styles including node-and-arc graphs to
  represent the data structure
• Programming interfaces.
• Administrative interfaces.

8
(No Transcript)
9
Logical data model

• A data model is a set of constructs for
  describing data
• The logical data model describes the structure
  data at the conceptual layer.
• To create a database, its structure must be
  specified using the modeling constructs
  available. Such a specification is called a
  schema.

10
Transaction

• A transaction consists of one or more changes to
  the database which take the database which take
  the database from one consistent state to
  another.
• Properties of transaction
• Atomicity. If a transaction is made up of a
  number of component updates, then every update
  must take effect or none of them should. The sum
  of money must not be debited without being
  credited or vice versa.
• Consistency. The total effect of all of the
  components of the transaction must take the
  database from one consistent state to another.
  The total sum of money represented in the
  database must remain unchanged.
• Isolation. Each transaction must not affect the
  smooth and correct running of any other
  transaction. Thus no two transactions are allowed
  to try to change the same account at the same
  time, for fear of corrupting each other.
• Durability. After the transaction has
  successfully completed, nothing will permit the
  changes made to be lost - not even a system crash

11
Record-based DMBS

• Network DBMS
• Relational DBMS

12
(No Transcript)
13
(No Transcript)
14
New applications areas

• CAD (computer aided design) CAE(computer aided
  engineering) CIM(computer integrated
  manufacturing)
• CASE(computer aided software emgineering)
• OIS(Office information system)
• GIS(geographic information system)
• WWW

15
Shortcomings of record-based DBMS

• No direct support for
• Hierarchies (part-of or is-a).
• Advanced data types (spatial, multimedia).

16
ER model for relational database
17
EER involves inheritance
18
Derived data type TAXIS

• PERSON_CLASS PERSON with
• keys person_id (name, address)
• characteristics name PERSON_NAME
• address
  ADDRESS_VALUE
• phone
  PHONE_VALUE
• attribute_properties age AGE_VALUE
• sex
  SEX_VALUE
• end

19
Object Databases

• Motivation to overcome weaknesses of relational
  approach
• Richer data models.
• Closer integration with programming languages.

20
Kinds of object database

• Object relational (Oracle, DB2, PostgreSQL).
• Programming language centred (Objectivity,
  FastObjects, Versant, ObjectStore).

21
Object Database Standards

• SQL-99(SQL-3)
• Mainstream SQL standard.
• Oracle, DB2, Informix.
• Object data management group (ODMG)
• Outgoing mainstream object database standard.
• Objectivity, FastObjects, Versant.
• Java data objects (JDO)
• Java-specific object-based database access.
• Poet, Versant, ObjectStore, SolarMetric,

22
A Matrix for classifying database application
23
Quadrant1simple data without queries

• cases text editor, VOD (video on demand)
• Database simple file system

24
Quadrant2simple data with queries

• database relational database
• standard SQL-99

25
Quadrant2simple data with queries

• Create table dept (
• dname varchar(20,
• budget float,
• floor int)

• create table emp (
• name varchar (30),
• startdate date,
• salary float,
• dept varchar(20))

• Find the names of employees with an employment
  start date after 1980 and who earn more than
  40,000
• Select name from emp where startdategt
  1980-12-31 and salary gt 40000
• Find the names of employees who work on the 1st
  floor
• Select name from emp where dept in
• (select dname from dept where floor1)
• Find the average salary of employees in the
  marketing department.
• Select average(salary) from emp where
  deptmarketing

26
Quadrant2simple data with queries

• Comments
• Client tool Client tool (i.e. 4GL) to set up
  forms for data entry and display is required.
• Performance transaction mechanism two-phase
  locking write-ahead log reduces performance.
• Security/architecture
• Market crowded

27
Quadrant3complex data without queries

• database object-oriented database
• standard ODMG, JDO

28
Quadrant3complex data without queries
create table emp ( name varchar (30),
space polygon, adjacency set-of
(empolyee))
create table floors ( number int, asf
swiss_cheese_polygon)

• Objective arrange employees onto floors

main() read all employees read all
floors compact() write all
employees
main() compact()
29
Quadrant3complex data without queries

• Comments
• Query language is not required.
• Client tools O-venders are expected to support
  many functions like compact().
• Performance persistent object will slow down
  performance. The user wishes to run no more than,
  say, 10 shower than the nonpersistent case.
• Security/architecture
• Market ltlt relational DB.

30
Quadrant4complex data with queries

• database object-relational database (ORDB)
• standard SQL-99

31
Quadrant4complex data with queries

• Example
• The state of California Department of Water
  Resource (DWR) manages most of the waterways and
  irrigation canals in California, as well as a
  collection of aqueducts. To document their
  facilities, DWR maintains a library of 35-mm
  slides. Over time this library has grown to
  500,000 slides and accessed many times a day by
  DWR employees and others.
• Problem clients usually requests a picture by
  content, ex. sunset. But it is difficult to
  find slides by content.

32
Quadrant4complex data with queries

• Create table slides (
• id int,
• date date,
• caption document,
• picture photo_CD_image)
• Create table landmarks (
• name varchar(30),
• location point)

• Find the sunset picture taken within 20 miles of
  Sacramento
• Select id from slides P, landmarks L, landmarks S
• where sunset (P.picture) and contains
  (P.caption, L. name) and
• L.location S.location and S.nameSacramento
  

33
(No Transcript)
34
Quadrant4complex data with queries

• Comments
• Query language that allows user-defined functions
  and operations (SQL-99).
• Client tools Standard 4GLs are useless. Advanced
  client tools are required. Ex. The case of DWR
  requires Zoom in/out user tool.
• Performance user-defined function might
  time-consuming. Query optimizer is required. Ex
• where sunset(image) and date lt 1985-01-01
• Security/Architecture client-server
• Market