S.sullivan DBMS

8/4/2019 S.sullivan DBMS

1/18

OODBMS, MDDBMS

Two emerging (emerged?)database technologies, and howthey relate to Knowledge

Management

Sean Sullivan

100046263


2/18

Outline

Issues with RDBMS

OODBMS

Benefits, drawbacks

MDDBMS

Benefits, drawbacks

OLAP, MOLAP

Summary

Questions


3/18

Relational Databases

Have problems with redundancy

Could normalize the data to create a starschema, but this breaks the objectacross multiple tables, and can introduceartificial keys

Complex queries are difficult or

sometimes impossible to write becauseSQL breaks the relational model

Traditionally, could not represent filessuch as pictures, audio and video

natively (only BLOB)


4/18

OODBMS: A Solution?

Started in the 1980s

Integrates database capabilities with OOprogramming language capabilities

Makes database objects appear asprogramming language objects in an existinglanguage (such as JAVA, C++ or Smalltalk)

Relational Model is no longer valid: objects arerelated through inheritance and polymorphism

More of a persistent programming languagethan a DBMS


5/18

Why Object Oriented?


6/18

Benefits of OODBMS

Explicit relationships improve the data accessperformance (especially as the database andcomplexity of the relationships grow) when

compared to value based relationships. Supports a large number of different types of

data, relationships, and objects with complexbehavior

A good fit for Knowledge Managementproblems, which are inherently complex

Found application in telecommunications, highenergy physics and subsets of financial services


7/18

The drawbacks

No clear standard

(ODMG didcome up with one, but it was not acceptedby the major OODBMS vendors)

Lack of interoperability with traditional RDBMStools (such as OLAP, backup and recovery,reporting, etc)

Seen by some as a lost opportunity to

revolutionize software and databasedevelopment

Today, there are more products that performobject-sql maping than there are pure OODBMS


8/18

Relational Database (revisited)

Represents data in a table

Think of displaying this information

in a spreadsheet Requires indexing and sorting which

takes time

SQL may not be able to get theinformation we need in a complexquery


9/18

MDDBMS

Multidimensional data model emergedover the past 10-15 years

MDDBMS is the Rubik's Cube of

database management systems Focuses on analyzing the data, not

recording transactions

Data is categorized as either facts withnumerical measures, or as dimensionsthat characterize the fact


10/18

Example

I decide to buy an AMD 64 3500+processor from http://tigerdirect.ca

The purchase is a fact; the priceand amount purchased arenumerical measures

The type of product I bought, whenI bought it, and where I bought itare all dimensions.


11/18

MDDBMS cont

Takes data from many sources, such asRDBMS, Legacy System, etc

Data is physically stored on disk in adata structure that is highly optimizedfor multidimensional processing and fastretrieval

Storage is between 2 and 10 times moreefficient over RDBMS due to betterindexing, compression andrepresentation of sparse data


12/18

MDDBMS = (M)OLAP?

Similar, but not the same

MDDBMS is a database, and evolveswith time as new data trickles in

OLAP is a snapshot

Both provide a multidimensional view ofthe data

MOLAP tends to generate faster queriesand uses less space than ROLAP orHOLAP


13/18

Benefits

Queries are simply a request to seepre-existing data organized in aspecific fashion.

Already highly organized, so therequested data is removed andreorganized

Stores information in the same waythat it is viewed (less datamanagement, and maintenance)


14/18

The drawbacks

Not the best solution for everyproblem

Works only on information withinterrelations

Database explosion with largeamounts of sparse data (calculating

all relationships can increase thedatabase size dramatically).


15/18

Summary

OODBMS has little to do with KM (infact, there is just plan very little

work being done in this field). MDDBMS are an important tool in

KM, and, although are not perfect,will be around for a while


16/18

Links of Interest

Object persistence in Java: Java DataObjects (JDO)

http://java.sun.com/products/jdo/

The OLAP Reporthttp://www.olapreport.com/


17/18

References

Colliat, G. (1996). OLAP, relational,and multidimensional database

systems ACM Press, 25(3), 64-69. http://wikkipedia.org

http://olapreport.org

http://cs.cmu.edu/People/clamen/OODBMS/Manifesto/htManifesto/Manifesto.html


18/18

Any

Questions, comments, queries,concerns (or quotes)?

S.sullivan DBMS

Documents