Management of the Evolution of Database-Centric Information Systems

1
Management of the Evolution of Database-Centric
Information Systems
Panos Vassiliadis2 , George Papastefanatos1,
Timos Sellis1, Yannis Vassiliou1 1 National
Technical University of Athens-KDBMS
Laboratory gpapas,timos,yv_at_dblab.ece.ntua.gr 2
University of Ioannina pvassil_at_cs.uoi.gr
2
Database Schema Evolution

• Current database systems are continuously
  evolving environments, where design constructs
  are
• Added
• Removed
• Modified
• Evolution is not handled by current DBMS with an
  automatic way
• E.g., syntactic as well as semantic adaptation of
  queries and views time-consuming task, treated
  in most of the cases manually
• Evolution driven database design is missing

3
Issues in Database Schema Evolution
4
Empirical Studies on Database Schema Evolution

• Collect test cases to come up with fundamental
  laws that govern database evolution
• Establish an experimental protocol for monitoring
  existing real-world databases and discover the
  way they evolve
• Collect such results and make them available to
  the research community

5
Principled Description of the architecture of a
database-centric Information System

• A commonly agreed formalism to express the
  internals of a database-centric system
• a model of the structural properties and dynamics
  of database-centric systems
• interdependencies in the components of a complex
  database-centric system
• a scientific foundation for the architecture of
  complex database-centric systems

6
Principled Response to evolutionary events

• Forecast the impact of an evolution event
• Handle conflicts
• Keep versions of the database consistent to user
  views
• Treat evolution in the presence or absence of
  user requirements
• Automation of the reaction to changes
• self-monitoring
• impact prediction
• auto-regulation
• self-repairing

7
Metrics

• Measures for the quality of the design of a
  database wrt its evolution
• Right families of such measures

8
Design Patterns

• Come up with a well-founded theory for design
  patterns wrt evolution
• Eliminate maintenance traps due to ad-hoc, or
  complex solutions
• Methods for testing database designs

9
Database Schema Evolution Our approach
Graph based representation of database constructs
(i.e., relations, views, constraints, queries)
Mechanism for performing what-if analysis for
potential changes of database configurations
Annotation of graph with rules for adapting
queries to database schema evolution
10
Graph based representation
Q SELECT EMP.Emp, Sum(WORKS.Hours) as
T_Hours FROM EMP, WORKS WHERE EMP.Emp
WORKS.Emp GROUP BY EMP.Emp
11
Annotation of graph with rules for adapting
queries to database schema evolution

• Set of evolving database constructs
• relations
• attributes
• constraints

• Set of reaction policies
• propagate
• block
• prompt

• Set of potential evolution changes
• addition
• deletion
• modification

According to prevailing policy, the proper action
is taken ? graph transformation
12
An Example
13
Conclusions Future Work

• A tool for visualizing and performing what-if
  analysis for several evolution scenarios.
• SQL extensions for annotating graph constructs
  with evolution semantics
• Evolution-driven database design
• Design patterns for allowing smooth evolution
• Definition of graph based metrics evaluating the
  system design