An Introduction to Object and Post-Relational Databases

1
An Introduction to Object and Post-Relational
Databases

• Presented by Jeff Moser

2
The Evolution of Object Technology

• Why objects?
• Objects closely represent human thinking.
• Hierarchical representation of structures.
• Easier to learn more productive
• C and Java pave the way
• C extends the market standard C language
• Java a free language the internet loves!

3
Relational Recap

• As soon as there was programs there was data
• Relational Databases
• Store data in flat tables
• Data management based on set theory
• Identifies relationships using keys
• Require normalized structure for optimal
  performace

4
The success of relational DBs

• Originated as a solution for storing and
  retrieving flat lists.
• Huge market support
• Standardized UI method (SQL)
• Many choices commercial and open source
• Redundancy and Reliability

5
The success of OO programming

• No longer a fad but the standard!
• Many Languages same concepts.
• High Schools and Colleges Agree
• How many of you learned an OO language in High
  School?

6
So where do I put my data?

• The four most common models
• Relational Databases
• Object-Relational Databases
• Object Databases
• Post-Relational Databases (also referred to as
  transactional multidimensional)

7
Relational Databases

• Relational DBs not designed to natively handle
  objects.
• O-R mapping
• Lose inheritance and relationships
• Added design time
• Deconstruction / Reconstruction of Objects
• The garage example
• The impedance mismatch

8
Object-Relational Databases

• Many different flavors
• Store complex objects in columns.
• Utilize stored procedures and caches to speed
  queries and access objects.
• Store tables in tables (Oracle Cartridges).
• Not a solution, but a band-aid.

9
Object Databases

• Stores all data in an object model
• Nearly infinite flexibility in defining and
  storing object data.
• Object model is identical for DB and Application.
• WAIT!! This is a bad thing!
• If the application changes so must the database.
• The database cannot be shared!

10
Post-Relational Databases

• Simultaneous SQL and object access
• Data dictionaries define relationships
• OO platform independent shared languages
• UML support for database engineering
• Easier to maintain
• Faster out of the box?
• Easier to learn?

11
Post-Relational Players

• Matisse www.matisse.com
• InterSystems Cache www.intersystems.com
• Both offer free developers editions.
• Both have extensive documentation available
  online.
• Both are cross-platform and cross-development
  environment.

12
An OO vs. Relational Example

• Consider the Car Registration Example Below

13
An OO vs. Relational Example

• Now Consider the Relational Tables

14
Comparing Access - Cache

• SQL and Cache Object Script

15
Comparing Access - Matisse

• A simple e-commerce application

16
Comparing Access - Matisse

• O-R mapping adds a lot of ugly code in order to
  construct / destruct objects that are stored or
  retrieved from the database.

17
Comparing Access - Matisse

• Storing objects natively in the database makes a
  lot more sense.

18
Comparing Access - Matisse

• Retrieving objects natively makes even more
  sense.
• Here we see a standard SQL query used to return
  an object rather than a table.

19
So why are RDBMS still the standard?

• People didnt start using C right away either!
• Large market share proven reliability
• At the beginning of this year the database market
  looked like this
• Oracle 41.3
• IBM 30.6
• Microsoft 13.4 - Windows Only
• Sybase 3.1
• Teradata 2.6
• All Others 9 - This includes mySQL, other open
  source dbs and all other databases.

20
So why are RDBMS still the standard?

• Key players focused on the competition.
• ODBMS and PRDBMS a cult following!
• O-R options and tools improving
• Oracle Cartridges
• OO Performance boosts countered with better
  Relational Access
• Standardization everyone loves SQL

21
QA

• For more information