Lab 4: Distribution Design

1
Lab 4 Distribution Design

• COMP334 Distributed Databases Systems
• CS Department, HKUST
• Wenwu LOU, Daniel

2
Contents

• Distribution Design
• Design Criteria
• QA for Course Project Phase I
• Get Started with JDBC for Oracle

3
Data Fragmentation

• Identify links between relations
• The direction of a link shows the one-to-many
  relationship
• Decisions
• For each owner relation, determine
• Whether you need to perform primary fragmentation
  
• If yes, how?
• For each member relations, perform derived
  fragmentations if necessary.
• Checking for Correctness
• Completeness
• Reconstruction
• Disjoint-ness

4
Data Allocation

• Collection Information
• Database information
• Application information
• Site information and Network informaiton
• Decisions
• Distribute relations and relation segments.
• Cost
• Performance

5
In Your Design

• Fragmentation Model
• Specify fragmentation model ONLY for relations
  that are to be fragmented.
• Follow the format in the textbook
• Allocation Model
• It is NOT required to follow the algorithms
  strictly.
• Give intuitive justifications

6
Design Criteria - Completeness

• Is the schema complete with respect to
  requirements?
• Are the requirements complete with respect to the
  schema?
• Make sure your design contains the following 4
  parts
• E/R diagrams
• Identify all the necessary entities and
  relationships, and their properties in the system
  that need to be stored in databases.
• Relational model
• Should be consistent to your E/R diagram
• Normalized to 3NF
• Fragmentation Schema
• Allocation Schema

7
Design Criteria - Correctness

• Your E/R model is precise.
• For example, you should not confuse entities with
  relationships.
• Your relational models must be consistent to your
  E/R model.
• Your fragmentation schema should be complete,
  reconstructable and disjointed.
• Your allocation schema fits the application
  requirement.

8
Design Criteria Conciseness and Justification

• Your design should be as concise as necessary.
• No redundancy in your E/R model and relational
  model
• Your design should be as expressive as it should
  be.
• No ambiguous.
• Justify your design whenever necessary.
• Try to convince me that your design is the best
  one

9
ER Diagram Annotations
Normal Attribute Derived Attribute Multi-value
attribute Composite Attribute Primary Key
Attribute Partial Key Attribute
Strong Entity Weak Entity Normal
Relationship Identifying Relationship
Cardinality (One) Cardinality (Many) Total
Participation
10
Common Mistakes for E/R model

• Fail to follow the predetermined notations.
• Missing identifier for an entity or relationship
• Wrong or unnecessary identifier for an entity or
  relationship
• Missing attributes
• Unnecessary or redundant attribute
• Confuse attributes with entities.
• Wrong attribute type (normal or derived)
• Missing, unnecessary, or redundant entities
• Wrong entity type (strong and weak)

11
More Common Mistakes

• Missing, unnecessary, or redundant relationships
• Missing or wrong mapping cardinalities in a
  relationship
• Missing or wrong participation constraints in a
  relationship
• Missing or wrong ISA hierarchy
• Missing or wrong covering and overlap constraints
  for ISA hierarchy.
• Using an entity instead of a relationship.
• Treating transactions as entities and
  relationships.

12
The Rest of the Lab

• QA for course project
• JDBC or Oracle