infs4201 Distributed Enterprise Computing

1
infs4201Distributed Enterprise Computing

• II. Distributed Information Management
• Dr Ralf Muhlberger
• ralf_at_itee.uq.edu.au

2
Overview

• Schema Integration
• Conflicts Dependencies
• Workflow Modelling
• Specifically data flow
• Transactions
• In relation to workflow

3
Schema Integration

• Fragmentation
• horizontal and/or vertical split
• potential replication
• Integration
• naming conflicts
• structural conflicts
• relationships between schemata

4
Naming Conflicts

• Synonyms
• P and PersonnelNo
• Homonyms
• Title (land ownership) vs Title (employment type)

5
Structural Conflicts

• Type conflicts
• Dependency conflicts
• e.g. one-to-one vs one-to-many
• Key conflicts
• Behavioural conflicts
• e.g. deleting last employee of a department

6
Relationships

• Between data possible
• Equivalent
• Subset
• Overlap
• Disjoint

7
Overview

• Schema Integration
• Conflicts Dependencies
• Workflow Modelling
• Specifically data flow
• Transactions
• In relation to workflow

8
Distributed Database

• Top Down
• Splitting a database across multiple sites
• Schema Fragmentation
• Based on each sites need for data
• Replication Management
• For data needed at more than one site

9
Multidatabase

• Bottom Up
• Start with pre-existing database systems
• Schema Integration
• Create a global schema/virtual database
• Autonomy and Performance
• MDBMS and local DBMS interaction

10
Workflow

• Top Down
• Business processes drive interoperability
• More design work
• Data structure and operations not as clear
• Less performance overhead
• Relaxing ACID properties

11
Business Process Specification

• Task - models a specific piece of work
• Input Sets with typed Inputs
• Output Sets with typed Outputs
• Possible Exception Outputs

Task Name
Review a Paper
Input
Output
Output Set
Input Set
12
Control Flows

• Define causal ordering (X must happen before Y
  can)
• Supports AND and OR semantics
• E.g. (X or Y) and Z

X
Y
T2
T1
Z
13
Data Flows

• Define causal ordering (X must happen before Y
  can) and carry data as payload
• Supports AND and OR semantics
• E.g. (p1 X or p1 Y) and p2 Z

X
Y
T2
p1
p2
T1
Z
14
Overview

• Schema Integration
• Conflicts Dependencies
• Workflow Modelling
• Specifically data flow
• Transactions
• In relation to workflow

15
ACID - Atomicity

• Atomicity
• The entire sequence of actions must be either
  completed or aborted. The transaction cannot be
  partially successful.
• Workflow
• Backward recovery in face of non-locking
  transactions

16
ACID - Consistency

• Consistency
• The transaction takes the resources from
  one consistent state to another.
• Workflow
• Broken down into individual tasks maintaining
  consistency.
• For interdatabase dependencies, can add
  consistency tasks into process model!

17
ACID Isolation

• Isolation
• A transaction's effect is not visible to other
  transactions until the transaction is committed.
• Workflow
• Usually ignored in workflow, as we have limited
  atomicity.

18
ACID - Durability

• Durability
• Changes made by the committed transaction are
  permanent and must survive system failure.
• Workflow
• Part of the workflow management system recovery
  features and also of the applications.