Third-Generation Database System manifesto

1
Third-Generation Database System manifesto

• The committee for advanced Database Management
  System (DBMS) Function

2
Question

• What is the path we must follow to
• arrive at the next generation of
• DBMS?

3
Current state description from authors (1990)

• Mainly 2nd generation DBMS
• Agreement in industry that the current DBMS need
  extended capabilities.
• There is a need for getting the common
  understanding of where we are going down on the
  paper.
• A counterweight to the spreading object-oriented
  enthusiasm is needed.

4
What is needed

• A gradual transition from 2nd
• generation DBMS to a 3rd
• generation DBMS

5
First Tenet

• Besides traditional database management
• services, third generation DBMSs will
• provide support for richer object structures
• and rules

6
Second Tenet

• Third generation DBMSs must subsume
• second generation DBMS.

7
Third Tenet

• Third generation DBMSs must be open to
• other subsystems.

8
(1. Tenet)Object and Rule Management

• A rich type system is needed (including
  extensions to query language).
• (Multiple- ) Inheritance is a good idea.
• Functions, including database procedures and
  methods, and encapsulation are a good idea.
• Unique IDs should only be reassigned if human
  defined primary key is not available.
• Rules are important.

9
(2. Tenet) Increasing DBMS Function

• Programmatic access should be through a
  non-procedural high-level access language
  (minimal navigation).
• Collections should be definable both through
  member enumeration and through queries
  (extensionally vs. intentionally).
• Updatable views are essential.
• Performance indicators must not appear in data
  models.

10
(3. Tenet)Need for an open system.

• DBMSs must be accessible through multiple higher
  level languages.
• Persistent objects is a good idea.
• SQL will prevail.
• Queries and result set should be the lowest level
  of communication.

11
Needed improvements in current DBMS to become 3rd
generation DBMS

• Improvements in relational databases systems
  inheritance and additional type constructors.
• Improvements in object-oriented database systems
  query language, query optimizers, views and SQL
  support.
• Both database systems need support for persistent
  programming languages.

12
Conclusion

• Object-oriented databases is not the solution to
  the challenges found in using 2nd generation
  DBMSs.
• The existing experience and expertise built into
  the existing relational databases should be used
  as much as possible.
• Development based on the current 2nd generation
  DBMS is more sensible than new development from
  scratch.