0% found this document useful (0 votes)
10 views60 pages

Distributeddbms Er. Inderjeet Bal

Uploaded by

shop.inderjeet
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
10 views60 pages

Distributeddbms Er. Inderjeet Bal

Uploaded by

shop.inderjeet
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 60

Distributed Databases

Er. Inderjeet Bal


Dept. of CS & IT
Concepts
Distributed Database.
A logically interrelated collection of shared
data (and a description of this data),
physically distributed over a computer
network.

Distributed DBMS.
Software system that permits the
management of the distributed database and
makes the distribution transparent to users.
Concepts
Collection of logically-related shared data.
Data split into fragments.
Fragments may be replicated.
Fragments/replicas allocated to sites.
Sites linked by a communications network.
Data at each site is under control of a DBMS.
DBMSs handle local applications autonomously.
Each DBMS participates in at least one global
application.
Distributed Processing and Distributed Databases
Distributed processing does not require a
distributed database, but a distributed
database requires distributed processing
Distributed processing may be based on a
single database located on a single computer
For the management of distributed data to
occur, copies or parts of the database
processing functions must be distributed to all
data storage sites
Both distributed processing and distributed
databases require a network of
interconnected components
Characteristics of Distributed Management Systems
Application interface to interact with the end user,
application programs and other DBMSs within the
distributed database
Validation to analyze data requests for syntax
correctness
Transformation to decompose complex requests into
atomic data request components
Query optimization to find the best access strategy
Mapping to determine the data location of local and
remote fragments
I/O interface to read or write data from or to
permanent local storage
Characteristics of Distributed Management Systems (cont’d.)

Formatting to prepare the data for presentation


to the end user or to an application
Security to provide data privacy at both local and
remote databases
Backup and recovery to ensure the availability
and recoverability of the database in case of
failure
DB administration features for the DBA
Concurrency control to manage simultaneous
data access and to ensure data consistency
across database fragments in the DDBMS
Characteristics of Distributed Management Systems
(cont’d.)

Transaction management to ensure the data


move from one consistent state to another
Must perform all the functions of centralized
DBMS
Must handle all necessary functions imposed
by distribution of data and processing
Must perform these additional functions
transparently to the end user
site 1
GDD
DDBMS

DC LDBMS

GDD

Computer Network
DDBMS

DC

site 2 DB

LDBMS : Local DBMS component


DC : Data communication component
GDD : Global Data Dictionary
The Ideal Situation
A single application should be able to operate
transparently on data that is:
spread across a variety of different DBMS's

running on a variety of different machines

supported by a variety of different operating


systems
connected together by a variety of different
communication networks
The distribution can be geographical or local
Workable definition
A distributed database system consists of a collection of
sites connected together via some kind of
communications network, in which :
Each site is a database system site in its own right;
The sites agree to work together, so that a user at
any site can access data anywhere in the network
exactly as if the data were all stored at the user's
own site
It is a logical union of real databases
It can be seen as a kind of partnership among individual
local DBMS's
Difference with remote access or distributed processing
systems
Temporary assumption: strict homogeneity
Distributed DBMS
Distributed Processing
A centralized database that can be accessed
over a computer network.
Parallel DBMS
A DBMS running across multiple processors
and disks designed to execute operations in
parallel, whenever possible, to improve
performance.
Based on premise that single processor systems
can no longer meet requirements for cost-
effective scalability, reliability, and
performance.
Parallel DBMSs link multiple, smaller machines
to achieve same throughput as single, larger
machine, with greater scalability and reliability.
Parallel DBMS
Main architectures for parallel DBMSs are:

a: Shared memory.


b: Shared disk.
c: Shared nothing.
Parallel DBMS

9
Advantages of DDBMSs
Organizational Structure
Shareability and Local Autonomy
Improved Availability
Improved Reliability
Improved Performance
Economics
Modular Growth
Disadvantages of DDBMSs
Complexity
Cost
Security
Integrity Control More Difficult
Lack of Standards
Lack of Experience
Database Design More Complex
Types of DDBMS
Homogeneous DDBMS

Heterogeneous DDBMS
Homogeneous DDBMS
All sites use same DBMS product.

Much easier to design and manage.

Approach provides incremental growth and

allows increased performance.


Homogeneous Database

Identical DBMSs
Heterogeneous DDBMS
Sites may run different DBMS products, with
possibly different underlying data models.
Occurs when sites have implemented their own
databases and integration is considered later.
Translations required to allow for:
Different hardware.
Different DBMS products.
Different hardware and different DBMS products.
Typical solution is to use gateways.
Heterogeneous Environment

Non-identical DBMSs
Open Database Access and Interoperability
Open Group has formed a Working Group to
provide specifications that will create database
infrastructure environment where there is:
Common SQL API that allows client applications to
be written that do not need to know vendor of
DBMS they are accessing.
 Common database protocol that enables DBMS from one
vendor to communicate directly with DBMS from another
vendor without the need for a gateway.
 A common network protocol that allows communications
between different DBMSs.
Most ambitious goal is to find a way to enable
transaction to span DBMSs from different vendors
without use of a gateway.
Multidatabase System (MDBS)
DDBMS in which each site maintains complete
autonomy.
DBMS that resides transparently on top of existing

database and file systems and presents a single


database to its users.
Allows users to access and share data without

requiring physical database integration.


Non-federated MDBS (no local users) and federated

MDBS (FMDBS).
Functions of a DDBMS
Expect DDBMS to have at least the

functionality of a DBMS.
Also to have following functionality:

 Extended communication services.


 Extended Data Dictionary.
 Distributed query processing.
 Extended concurrency control.
 Extended recovery services.
Reference Architecture for DDBMS
Due to diversity, no universally accepted architecture
such as the ANSI/SPARC 3-level architecture.
A reference architecture consists of:
 Set of global external schemas.

 Global conceptual schema (GCS).

 Fragmentation schema and allocation schema.

 Set of schemas for each local DBMS conforming to 3-level

ANSI/SPARC .
Some levels may be missing, depending on levels of
transparency supported.
Reference Architecture for MDBS
In DDBMS, GCS is union of all local conceptual

schemas.
In FMDBS, GCS is subset of local conceptual

schemas (LCS), consisting of data that each local


system agrees to share.
GCS of tightly coupled system involves integration

of either parts of LCSs or local external schemas.


FMDBS with no GCS is called loosely coupled.
Reference Architecture for Tightly-Coupled Federated
MDBS
Components of a DDBMS

25
Distributed Database Design
Three key issues:

Fragmentation.

Allocation

Replication
Distributed Database Design
Fragmentation
Relation may be divided into a number of sub-
relations, which are then distributed.

Allocation
Each fragment is stored at site with "optimal"
distribution.

Replication
Copy of fragment may be maintained at several
sites.
Fragmentation
Definition and allocation of fragments carried
out strategically to achieve:
Locality of Reference

Improved Reliability and Availability

Improved Performance

Balanced Storage Capacities and Costs

Minimal Communication Costs.

Involves analyzing most important applications,


based on quantitative/qualitative information.
Fragmentation
Quantitative information may include:

 frequency with which an application is run;

 site from which an application is run;

 performance criteria for transactions and


applications.

Qualitative information may include transactions

that are executed by application, type of access


(read or write), and predicates of read
operations.
Data Allocation
Four alternative strategies regarding
placement of data:
Centralized

Partitioned (or Fragmented)

Complete Replication

Selective Replication
Data Allocation
Centralized

 Consists of single database and DBMS stored at

one site with users distributed across the network.

Partitioned

 Database partitioned into disjoint fragments, each

fragment assigned to one site.


Data Allocation
Complete Replication

Consists of maintaining complete copy of

database at each site.

Selective Replication

Combination of partitioning, replication, and

centralization.
33
Why Fragment?
Usage

Applications work with views rather than entire

relations.

Efficiency

Data is stored close to where it is most frequently

used.
Data that is not needed by local applications is not

stored.
Why Fragment?
 Parallelism
 With fragments as unit of distribution, transaction can be

divided into several subqueries that operate on fragments.


 Security
 Data not required by local applications is not stored and so

not available to unauthorized users.

 Disadvantages
 Performance

 Integrity.
Correctness of Fragmentation
Three correctness rules:

Completeness

Reconstruction

Disjointness.
Correctness of Fragmentation
Completeness

If relation R is decomposed into fragments R1, R2, ...


Rn, each data item that can be found in R must
appear in at least one fragment.

Reconstruction

Must be possible to define a relational operation

that will reconstruct R from the fragments.


Reconstruction for horizontal fragmentation is

Union operation and Join for vertical .


Correctness of Fragmentation
Disjointness

If data item di appears in fragment Ri, then it should

not appear in any other fragment.


Exception: vertical fragmentation, where primary key

attributes must be repeated to allow reconstruction.


For horizontal fragmentation, data item is a tuple

For vertical fragmentation, data item is an attribute.


Types of Fragmentation
Four types of fragmentation:

 Horizontal
 Vertical
 Mixed
 Derived.

Other possibility is no fragmentation:

 If relation is small and not updated frequently, may


be better not to fragment relation.
41
Mixed Fragmentation
Horizontal Fragmentation
This strategy is determined by looking at
predicates used by transactions.
Involves finding set of minimal (complete and
relevant) predicates.
Set of predicates is complete, if and only if, any
two tuples in same fragment are referenced with
same probability by any application.
Predicate is relevant if there is at least one
application that accesses fragments differently.
Transparencies in a DDBMS
Distribution Transparency
Fragmentation Transparency
Location Transparency
Replication Transparency
Local Mapping Transparency
Naming Transparency
Transparencies in a DDBMS
Transaction Transparency

Concurrency Transparency
Failure Transparency

Performance Transparency

DBMS Transparency
Distribution Transparency
Distribution transparency allows user to
perceive database as single, logical entity.
If DDBMS exhibits distribution transparency,
user does not need to know:
data is fragmented (fragmentation
transparency),
location of data items (location transparency),
otherwise call this local mapping transparency.
With replication transparency, user is
unaware of replication of fragments .
Naming Transparency
Each item in a DDB must have a unique
name.
DDBMS must ensure that no two sites create
a database object with same name.
One solution is to create central name server.
However, this results in:
loss of some local autonomy;
central site may become a bottleneck;
low availability; if the central site fails,
remaining sites cannot create any new objects.
Transaction Transparency
Ensures that all distributed transactions
maintain distributed database’s integrity and
consistency.
Distributed transaction accesses data stored at
more than one location.
Each transaction is divided into number of sub-
transactions, one for each site that has to be
accessed.
DDBMS must ensure the indivisibility of both
the global transaction and each
subtransactions.
Concurrency Transparency
All transactions must execute independently
and be logically consistent with results
obtained if transactions executed one at a
time, in some arbitrary serial order.
Same fundamental principles as for
centralized DBMS.
DDBMS must ensure both global and local
transactions do not interfere with each other.
Similarly, DDBMS must ensure consistency of
all sub-transactions of global transaction.
Concurrency Transparency
Replication makes concurrency more
complex.
If a copy of a replicated data item is updated,
update must be propagated to all copies.
Could propagate changes as part of original
transaction, making it an atomic operation.
However, if one site holding copy is not
reachable, then transaction is delayed until
site is reachable.
Concurrency Transparency
Could limit update propagation to only those
sites currently available. Remaining sites
updated when they become available again.
Could allow updates to copies to happen
asynchronously, sometime after the original
update. Delay in regaining consistency may
range from a few seconds to several hours.
Failure Transparency
DDBMS must ensure atomicity and durability
of global transaction.
Means ensuring that sub-transactions of global
transaction either all commit or all abort.
Thus, DDBMS must synchronize global
transaction to ensure that all sub-transactions
have completed successfully before recording a
final COMMIT for global transaction.
Must do this in presence of site and network
failures.
Performance Transparency
DDBMS must perform as if it were a
centralized DBMS.
DDBMS should not suffer any performance
degradation due to distributed architecture.
DDBMS should determine most cost-effective
strategy to execute a request.
Performance Transparency
Distributed Query Processor (DQP) maps
data request into ordered sequence of
operations on local databases.
Must consider fragmentation, replication,
and allocation schemas.
DQP has to decide:
which fragment to access;
which copy of a fragment to use;
which location to use.
Performance Transparency
DQP produces execution strategy optimized
with respect to some cost function.
Typically, costs associated with a distributed
request include:

I/O cost;
CPU cost;
communication cost.
Thanks

You might also like