Wrapper-Based Evolution of Legacy Information System

1
Wrapper-Based Evolution of Legacy Information
System

• Philippe Thiran et al
• Fcculties University Notre-Dame de la Paix

2
Outline

• Why Data Wrappers?
• Building R/W Data Wrappers
• R/W Wrapper Architecture
• Generating Data Wrapper

3
Why Data Wrappers?

• Architectural shifts from centralized to
  distributed and cooperative paradigm
• Data consistency in legacy architecture
• Explicit constraints guaranteed legacy system
• Implicit constraints data consistency is
  guaranteed by client application code
• Example
• PR field of record type ORDER refer a definite
  PRODUCT record
• Implicit constraints to create or modify an
  ORDER, the referred PRODUCT should be first
  checked

ORDER PR
Product
4
Why Data Wrappers?

• Plug a new application component into such a
  legacy architecture
• How to validate the implicit data property?
• Forward Wrappers
• Translate data and queries from the legacy data
  model
• Take charge of the validation logic

5
Why Data Wrappers?

• Reuse a legacy component in a modern system
• Semantic mismatch many important constraints and
  structures are ignored by data management system
• Backward Wrappers
• Query and update the new data
• Data integrity is ensured by both the new and
  legacy app

6
What is a Data Wrapper?

• Component inserted between the legacy database
  and the application component to
• Extend the lifetime of existing system
• Address the heterogeneity by providing a standard
  and common interface
• Allows to transparently incorporate new
  capabilities
• Provides a smooth path to a step-by-step
  modernization of a complex legacy system
• Interface of a Data Wrapper
• A wrapper schema
• A common query language

7
Outline

• Why Data Wrappers?
• Building R/W Data Wrappers
• R/W Wrapper Architecture
• Generating Data Wrapper

8
Bridge the Semantic Gap

• Data Wrapper a new application component that
  query and update the contents of a legacy
  database
• Physical schema explicit structures and
  constraints
• Logical schema explicit and implicit structures,
  accessed by the wrapper
• By accessing the logical schema, data wrappers
  address the implicit constraints
• Reject the updates that violate the implicit
  constraints

9
Building R/W Wrapper

• Wrapper Schema Definition
• Recover explicit constraints
• Enriched with implicit constraints by reverse
  engineering techniques
• Program analysis
• Data analysis

10
Building R/W Wrapper

• Mapping definition
• Transform structures
• Transform constraints
• Example of update statement

11
Outline

• Why Data Wrappers?
• Building R/W Data Wrappers
• R/W Wrapper Architecture
• Generating Data Wrapper

12
R/W Wrapper Architecture

• Query/update analysis
• Error reporting
• Query/update and data translation
• Implicit constraint control
• Security
• Concurrency and transaction management

13
Specification of a Wrapper

• GER a wide spectrum entity-relationship model
• Generate sub-models
• Retain necessary parts in the sub-model
• Rename the constructs
• Assemble the valid parts
• Advantage
• All inter-model transformations equal to
  intra-model process
• A limit set of primitive operators is sufficient
  to model most database engineering processes
• Enriched Model
• Encompass the physical, logical and conceptual
  models
• Define a set of constructs comprising the
  structures and constraints

14
Model and Schema Specification

• Constructs entity types, attribute, relationship
  types
• Built-in Constraints

15
Transformational Mapping Spec

• Transform a schema from one model to another,
  requiring specific operators
• Inverse transformation
• T2(T1(C))C
• Transformation sequence
• S1-to-S2(T1 T2 ... Tn) S2T2(T1(S1))
• S2-to-S1(Tn-1 T2-1 T1-1) S1S2-to-S1(S2)
• Transformation categories
• T augment the semantics of the schema
• T Preserve the semantics of the schema
• T- Reduce the semantics of the schema
• Implicit constraints and schema Interpretation

16
Outline

• Why Data Wrappers?
• Building R/W Data Wrappers
• R/W Wrapper Architecture
• Generating Data Wrapper

17
Generating Wrapper

• Database reverse engineering
• Yield all the necessary information to specify
  and develop a wrapper
• Semi-hardcoded wrapper
• A model layer and a database layer
• Schema transformation-based wrapper generation
• T is used to specify the implicit constraint
  assertions instead of T
• Operational CASE support
• Schema processing, mapping processing, and
  wrapper generation

18
Database Reverse Engineering

• Results
• Wrapper schema
• Schema transformation sequence
• Process
• Physical extraction
• Logical extraction
• Wrapper schema derivation
• Mapping definition

19
Wrapper Generation
20
Wrapper Generation

• The results from DBRE
• Legacy physical schema
• Legacy logical schema
• Wrapper schema
• LS-to-PS mapping is formally defined, used to
  code the implicit constraint control component
• Structural part of WS-to-LS is used to code the
  query/update translator
• Instance part of ls-to-ws define the data
  translator

21
Reference

• Thiran, P. et al, Wrapper-Based Evoluation of
  Legacy Information System, ACM Tran Software
  Engineering and Methodology Vol.15 No.4, October
  2006
• Brodie, M. et al. 1995. Migrating Legacy Systems,
  Morgan Kaufmann
• Haninaut, J. et al, 2005, Transformation-Based
  Database Engineering, IDEA Group

22

• Thanks
• Questions?