Transcript Relational Databases - PersonalWebIWayanSW

Issues In Distributed Data
Management
Heterogeneous Distributed Database
- Philip K. Nukpe
1
Heterogeneous Distributed Database
Heterogeneous Distributed Database
- Philip K. Nukpe
2
Definition: ‘Composed of heterogeneous hardware, operating system
database management system and applications’
 Provides logically integrated view of existing heterogeneous
distributed databases.
 The three-level architecture of a Heterogeneous Distributed
Database System (HDDMS)
User
User
External
Schema 1
User
External
Schema 2
User
External
Schema 3
Conceptual
Schema
Internal
Schema
Heterogeneous Distributed Database
- Philip K. Nukpe
3
Challenges posed by integrating distributed
databases:
 Data management system (different vendors)
 Data Models (i.e. relational, text indexing, object)
 Query the transaction processing algorithms
 Data types (i.e., text graphics, multimedia, hypermedia)
 Format (i.e. Structured, unstructured)
Semantics
Heterogeneous Distributed Database
- Philip K. Nukpe
4
 In a heterogeneous database (HDD), the local database
is not managed by the same distributed database
management system.
 Federated Database:
“ A combination of autonomous, heterogeneous
databases that are operating together”
 Co-ordinating software is called Federated DBMS
 Component Databases are individual databases that take
part in a Federated Database System.
 Component databases can continue to execute local tasks
while participating in a federation.
Heterogeneous Distributed Database
- Philip K. Nukpe
5
 Typical Heterogeneous databases include:


World Wide Web
Multimedia
 Preservation of the database’s autonomy is paramount.
 Data access facilities in Heterogeneous Database
Systems can range from:
 browsing across component databases
 querying a centralised data warehouse
 querying multiple databases.
 A heterogeneous database system can be provided with
a multidatabase query language.
Heterogeneous Distributed Database
- Philip K. Nukpe
6
 This allows queries to be expressed with direct
reference to elements(data) of the component
database.
 Multidatabase has been classified according to the
following criteria:
1. Distribution:
Data can be placed in many distributed databases.
Databases can be found in the same or different
computer systems.
2. Heterogeneity:
This is found in the differences in technology i.e.,
software, hardware, and operating system.
Heterogeneous Distributed Database
- Philip K. Nukpe
7
The two types of heterogeneity are those that are due to:
- differences in various DBMSs
- differences in data semantics
3. Autonomy:
A component database can have the following autonomies:
• Design Autonomy (Choice of data presentation and
attributes)
• Communication Autonomy (ability to decide on
its own when to communicate with other
components of the federated database system.)
• Execution Autonomy:( ability to execute some of
its operations locally without influence by external
operations that are executed in other component
databases or in the federation database system)
Heterogeneous Distributed Database
- Philip K. Nukpe
8
 Component database in a federation although
autonomous, allow partial and controlled sharing of their
data.
A federated database can be classified into two classes
depending on who manages the federation and how the
component database are co ordinating:
 LOOSELY COUPLED FEDERATED DATABSE
SYSTEM



User required to take responsibility of producing and managing
the federation.
No control from the federation database or mangers.
Also called Interoperable
Database System
Heterogeneous Distributed Database
9
- Philip K. Nukpe
 TIGHTLY COUPLED FEDERATED DATABASE


Federate system and managers produce and manage the
federation
Responsible for controlling the component databases.
 PROBLEMS IN INTEGRATING
HETEROGENEOUS DATABASES

Semantic Heterogeneity (Different Database
designers represent the same object in different ways.)
e.g. Synonyms - same entity different names
Homonyms - different entities same names
Heterogeneous Distributed Database
- Philip K. Nukpe
10
 Description Heterogeneity(Different characteristics are
used to describe the same object)
 Model Heterogeneity(use of different models to represent
the same data e.g. SSADM, E-R Diagram, Use Cases)
REQUIREMENT FOR FDDMS
 Provision of a homogeneous interface (federated schema)
to overcome heterogeneity)
Autonomy of local databases(CDBS) to be maintained.
Federated schema to maintain definition of external schema.
The support of mechanism to guarantee consistency.
Heterogeneous Distributed Database
- Philip K. Nukpe
11
Characteristics of FDB Environment
 Due to decentralisation of corporate data,
heterogeneous software, hardware are required
hence FDBS is the natural choice for such
systems.
 Fragmentation Transparency
 Location Transparency
 Local Transparency
(Diagram )
Heterogeneous Distributed Database
- Philip K. Nukpe
12
External
Schema
External
Schema
Federated Schema
Export
Schema
External
Schema
Federated Schema
Export
Schema
Component
Schema
Local
Schema
LDBC
Export
Schema
Component
Schema
Local
Schema
LDBC
Heterogeneous Distributed Database
- Philip K. Nukpe
13
Federated Multidatabase Architecture
Local Schema
Conceptual Schema of a component Database
Component Schema
Translation of local Schema into data model
Export Schema
Subset of component Schema
Include access control
Heterogeneous Distributed Database
- Philip K. Nukpe
14
Federated Schema
integrates export Schema & Include information
on data distribution
support distribution feature of FDBS
External Schema
Define user/application interface
Heterogeneous Distributed Database
- Philip K. Nukpe
15
 Common (Canonical)Data Model (CDM)

mapping of a component data model database to the
data model of all participating systems in a
Multidatabase systems.
 Transaction Management
 Serialisability
Tickets
Heterogeneous Distributed Database
- Philip K. Nukpe
16
Tickets operations
• Only global transactions must take ticket values
• Local transaction don’t take ticket values
•Two Approaches to using Tickets
• Optimistic Ticket Method (OTM)
• Conservative (pessimistic) Ticket Method
(CTM)
Heterogeneous Distributed Database
- Philip K. Nukpe
17