Distributed Database SystemsAutumn, 2007

Chapter 4

Distributed DBMS Architecture

1Distributed Database Systems

Introduction

Architecture defines a system’s structure with◦ Components◦ Functions of components, and◦ Their interactions

Purpose of ‘reference architecture’:◦ A framework for discussion◦ Standardization

2Distributed Database Systems

4.1 Transparencies in a distributed DBMS

X transparency means the existence of X is not known to users.Closely related to architecture issues.

Distributed Database Systems 3

4.1.1 Data independence

Same as in centralized DBMS:◦ Logical data independence

applications are immune to changes in schema

◦ Physical independenceapplications are immune to changes in physical storage, and data distribution

Distributed Database Systems 4

4.1.2 Network transparencyAlso known as “distribution transparency”

4.1.3 Replication transparency

4.1.4 Fragmentation transparencyData fragmentation – the division of relation into smaller fragments at different sites.

Distributed Database Systems 5

4.2 DBMS Standardization

Distributed Database Systems 6

4.2.1 ANSI/APARC ArchitectureA 3-level architecture based on data organization

Distributed Database Systems 7

View Description

External View • How users view the shared data• Commands used:create view … as …

Conceptual View • Synthesis of external views of an enterprise• Commands used:create db …create table …

Internal View • Physical definition and organization of data• Commands used:index on devicesegment definition and allocation

The ASNI/SPARC Architecture (1997)

Distributed Database Systems 8

External View

External View

Ex ternal View

ConceptualView

Internal View

Ex ternal Schema

Conceptual Schema

Intern al Schema

Users

4.3 Architecture Models for Distributed DDBMS

Distributed Database Systems 9

4.3 Architecture Models for Distributed DDBMS

Classification of multiple DBs shared by multiple DBMSsCriteria for classification◦ Distribution

0: centralized, 1: C/S, 2: P2P◦ Autonomy

0: tight integration, 1: semi-autonomous, 2: total isolation

◦ Heterogeneity0: homogeneous, 1: heterogeneous

Distributed Database Systems 10

Distributed Database Systems 11

Distribution

Autonomy

Heterogeneity

1. (A0,D2,H0)P2P homogeneoustightly integrated

P2P homogeneousfederated

P2P homogeneousmultidatabase

2. P2P heterogeneousfederated

3. (A2,D2,H1)P2P heterogeneousmultidatabase

Note 1: the available most advanced commercial product 2 and 3: active research area

Autonomy: the distribution of control

(0) tightly integrated◦ a single image of the entire DB to and user

and a single DBMS(1) semi-autonomous◦ all DBMSs can operated independently but

have to join a federation to make local data sharable

(2) total isolation◦ each DBMS is stand-alone and enjoy 100%

freedom of autonomy

Distributed Database Systems 12

Distinction between DDB and MDB

In MDB◦ every local data manager is a complete DBMS

with its own DDL, DML and TM

In DDB◦ there is only one data manager with its

dependent component-data-manager at every local site

Distributed Database Systems 13

The focus of this course is the PRINCIPLES of DDBs (A0, D2, H0)

4.3.1 Distributed DBMS architecture

An architecture based on data organization

Distributed Database Systems 14

External SchemaExternal Schema External Schema

Global ConceptualSchema

Local Internal Schema

Local ConceptualSchema

Local ConceptualSchema

Local ConceptualSchema

Local Internal Schema

Local Internal Schema

ES

LCS

LIS

GCS

Detailed functional description of DDBMS

Legend

◦ Administrative Role

◦ Data Dictionary/Directory

◦ Processing Unit

Distributed Database Systems 15

Distributed Database Systems 16

Enterprise Administrator

Global DB Administrator

ApplicationSystem

Administrator

GCS Processor

ES Processor

Local Data ManagerLocal Data Manager

Local DBA

GCS Processor-1

LD/D Local IS

Processor

Internal DB

Conceptual Transformation

Int. StorageInt. DB

Transformation

……

……

GD/D

……

Components of a Distributed DBMS

Distributed Database Systems 17

Semantic data controller

• Integrity constraint and authorization

Global Query Processor

• decompose and optimize global query

Global Execution Monitor

• Distributed transaction manager

Runtime Support Processor

• physical access to database

Client/Server Systems

A computing model dividing functions between client and server

Distributed Database Systems 18

Discussion

Why DDB?Compare DDB with CDB, MDB, C/S DB.Do you understand DDB architecture?

Distributed Database Systems 19