Some references on distributed databases

UNIT TITLE: ADVANCED DATABASES

LEVEL: M

LECTURER Dr Yuan Sun

AIMS

To expose the student to the inherent flaws that exist in centralised relational databases. The student will then study those aspects of database design and implementation that attempt to alleviate these short comings.

At the end of the unit the student can confidently expect to be able to undertake the following;

• a critical appraisal of an existing system;
• design and implement a distributed database including predictions of its performance;
• design an object oriented database;
• a principle approach to the database performance issues.
  

RATIONALE

Although centralised relational databases represented a considerable improvement over hierarchical and networked models they are less than ideal for many applications.

Centralised databases are inappropriate today simply because of the migration from centralised systems and the growth of distributed systems. Numerous examples can be found in both the press and the literature. The clearing banks in the UK have traditionally centralised their operations with the raw material of transactions (cheques, etc.) being either physically or electronically transmitted to a central site for processing. These organisations are now moving towards local processing in the branch. Ultimately, the advantages of transparency will overtake the ad-hoc view of the various databases and the centralised database together with the branch databases will be seen as one distributed database.

OUTLINE SYLLABUS

Object Oriented Databases

• Semantic data modelling in database environment.
• The object oriented extended functional data model.
• Issues of persistent storage of entities add functions.
• Object oriented databases with multiple inheritance and meta classes.
  

Distributed Databases

• Design aspects. Methods of distribution: partitions, replication, partial replication.
• Networked data dictionaries and data management systems.
• Query processing, transaction switching/splitting.
• Distributed databases administration.
• Mechanisms for distributed control, failure, synchronisation, transaction structure.
• Performance issues.
  

Current Trends in Databases Development

• Databases for massively parallel systems.
• Distributed knowledge bases.
• Data mining.
• Temporal & spatial database.
  

LEARNING OUTCOMES

• At the end of successful completion of this unit a student will be able to :
• Design a distributed database.
• Implement a distributed database.
• Design an object oriented database.
• Critically appraise of a existing systems.
• Evaluate and predict the performance of a database system.
• Discuss the development of database technology.
  

UNIT ORGANISATION AND STRUCTURE

Students taking this unit will already have undertaken an extensive study of databases such as that found on an information system oriented computing first degree. The previous study implies that the student is in some sense already a computer professional and can be expected to see the material presented here against a backdrop of their experience.

A traditional lecture program will present the same material of each section using no more than 50% of the available time. The remaining 50% will be taken up with small scale practical and a major case study in each section in which the student will be expected to design and implement a sophisticated database using each of the paradigms.

LENGTH OF UNIT: 1 Semester

DATE UNIT STARTS: Sept. 1996 DATE UNIT ENDS: Feb. 1997

TIME SPENT EACH WEEK: Contact time: 4 hr. Private study: 4 hr.

ORGANISATION OF WEEKLY TIME SLOTS

In general the 4 hour block will be divided into three 1 hour lectures and 1 hour of student-based work in tutorials, seminars, group work and practical sessions( which will be carried out in the computer labs).

WEEK-BY-WEEK PLAN OF TOPICS

The following represents a guide to the lecture program, the weekly schedule may change if deemed necessary.

Week	Lecture Program 1 Topics	Lecture Program 2 Topics
1	Overview of DDBMS and computing network.	Lab work for OODB
2	Distributed DBMS architecture & design.	Lab work for OODB
3	Query optimisation	Lab work for OODB
4	Distributed query processing.	Lab work for OODB
5	Management of distributed transaction.	Lab work for OODB
6	Concurrency control & reliability.	Lab work for OODB
7	Distributed database administration.	Lab work for OODB
8	Data warehouse approach	Lab work for OODB
9	Temporal databases	Lab work for OODB
10	Parallel databases	Case studies.
11	Other database issues.	Research issues.
12	Workshop on object database	Workshop on object database
13	Revision/Slippage

ASSESSMENT

Emphasis will be placed on learning through active student participation in lectures and through dedicated practical sessions. The evaluation of effectiveness of all these aspects will be an important, recurring theme throughout the course.

The end-of-unit mark will be determined in the following way:

Course work 70%

Examination 30%

Course work schedule: Hand-out data: week 1 Hand-in date: week 11 (Tue. 4:00pm)

ASSESSMENT CRITERIA

Marks are awarded on the following basis:

A. (Over 70% and a distinction) represents comprehensive and competent coverage of issues, the correct use of more advanced techniques plus evidence of further reading and insight into the problem.

B. (60-70%) represents an overall competence and a willingness to tackle more advanced aspects of the problem.

C. (50-60%) represents an average attempt at a solution, mostly competent and not seriously lacking in any aspect and showing some reflective thinking.

D. (40-50%) represents a bare pass, an attempt to address all aspects of the problem but failing in one or more areas to achieve a minimal level of understanding.

E. (less than 40% and a failure) reflects a lack of serious attempt at or understanding of most of the key issues involved in the assignment, very late submission or very poor presentation.

COURSE WORK

Sarawak Transport Co. operates 200 bus/coach routes from 10 separate depots and 8 rail lines servicing 60 separate stations with 20 separate depots spread throughout the country. Some buses/coaches are OPO (one person operated) whereas others have both a driver and a conductor. All rail trains have a driver and a guard.

Your task is to design an integrated information system for passengers and staff. All the buses and trains have transponders which can inform the system via satellite where they are at any time. The eventual objective is to provide an evenly spaced service as possible for all its customers. You can make any extra assumptions you wish but you must document and justify them.

1. Discuss the choice of both distributed and centralised alternative implementations of this system.
2. Design an object-oriented database based on a centralised computer for the Sarawak Transport Co.
3. Implement your OODB schema by using GemStone.
4. Construct a query structure (or methods) for database interface.
5. Make a comparison between the relational (with distributed function) and object-oriented database approaches based on the Sarawak Transport Co.
6. Conclude your design and indicate the future work for improvements.

Your submission will be assessed on its innovation, originality and practicability.

EXAM FORMAT INSTRUCTIONS

LENGTH: 2 hours QUESTIONS: 4 from 6

All questions carry equal marks.

REFERENCES

Principal Source:

M. T. Ozsu & P. Valduriez Principles of Distributed Database Systems

Prentice-Hall International Editions, 1991.

Wom Kim Modern Database Systems, Addison Wesley, 1995.

Further Reading:

Bell & Grimson Distributed Databases Systems, Addison-Wesley, 1992.

J. G. Hughes Object-Oriented Databases, Prentice-Hall International, 1991.

D. E. Shasha Database Tuning -- A principled Approach, Prentice-Hall, 1992.

Ceri & Pelagatti Distributed Databases, McGraw-Hill, 1984.

C. J. Date An Introduction to Database Systems, Vol. I & II, 5th Edition

Addison-Wesley, 1990.

Gray, Kulkarni & Paton Object-Oriented Databases, Prentice-Hall, 1992.

Brown Object-Oriented Databases, McGraw-Hill, 1991.

R. G. G. Cattell Object Data Management -- Object-oriented and Extended Relational Database Systems, Addison-Wesley, 1991

H F Korth & A Silberschatz Database System Concepts, McGRAW-HILL, 1991.