Please enable JavaScript.
Coggle requires JavaScript to display documents.
Client-Server and Distributed Data Bases - Coggle Diagram
Client-Server and Distributed Data Bases
Motivation for Client-Server Processing and Distributed Data
The client-server
remote computing resources
a variety of subtasks
A client
a program that makes requests to a server
The server
performs requests and communicates results to the clients
arranged across networked computers
to divide the work
advantages
flexibility
scalability
interoperability
data control, communication costs and performance
organization's structure
sharing and maintaining data
Distributed data
80 percent of the requests are local.
Local requests incur little or no communication costs
Increased data availability
no single computer responsible for controlling access
data can be replicated
very difficult
Client-Server Database Architectures
Two-Tier Architecture
The database server processes the SQL statements
The PC client contains the presentation code
the database server performs process management functions
code can be split
client can invoke stored procedures
fat clients
good approach for systems with stable requirements
a moderate number of clients
the simplest to implement
software maintenance can be difficult
Open Database Connectivity (ODBC)
Three-Tier Architecture
a transaction-processing monitor
message-oriented middleware
a middleware server
application server
the additional server software
addresses performance limitations of the two-tier architecture
not address division-of-processing concerns
contain the same division of code
provide more flexibility on division of processing
Multiple-Tier Architecture
provide flexible division of processing
support additional layers of servers
improve performance
Web Services Architecture
deploying services faster
communicating new services
finding existing services
high interoperability
interaction between a service provider, service requestor, and service registry
The service provider
The service requestor
The service registry
Distributed Database Architecture
Schema Architectures
additional layers
fragmentation and allocation
definition of each fragment
vertical subset - project operation
horizontal subset - restrict operation
mixed fragment (combination of project and restrict operations)
allocated to one site
one copy of a fragment is considered the primary copy
Only the primary copy is guaranteed to be current
more autonomy
the traditional three schema levels
a local mapping schema
the exportable data
provide conversion rules
a global format
The global conceptual schema
all of the kinds of data
relationships
global requests
global external schemas provide views of shared data in a common format
differences among the local data formats
different DBMSs
different set of data types
The data models of the local DBMSs
legacy systems
file interfaces
navigational data models
do not support SQL
different data types, scales, units of measure, and codes
The local mapping schemas
conversion rules
data from a local format into a global format
The tightly integrated and loosely integrated architectures
extreme possibilities
proposed and implemented
provide additional local autonomy
more efficiency for global requests
can be combined
can be loosely integrated
share selective data
a gateway between tightly integrated distributed databases
Distributed Query Processing
more complex
involves both local (intrasite) and global (intersite) optimization
data movement
centralized query processing
a join
one fragment can be moved
both fragments can be moved to a third site
just the join values of one fragment can be moved
a site for each fragment
parallel databases
shared nothing architectures
much faster
more reliable communication networks
parallel database processing
reduce response time
increase the overall amount of resources consumed
multiple optimization objectives exist
centralized environment
minimizing resource
minimizing response time
distributed environment
minimizing resources may conflict
minimizing response time
parallel processing opportunities
the weighting of communication costs versus local costs depends on network characteristics
communication costs can dominate local costs
Transactions obey the ACID properties
the distributed DBMS provides concurrency and recovery transparency
the implementation of the principles more difficult
be coordinated
new kinds of failures
communication network
new protocols are necessary
Distributed DBMSs
support global requests
A site
locally controlled computer
unique network address
geographically distributed
located in close proximity
Global requests
queries
combine data
update data
can involve a collection of statements
accessing local data
remote data
Local data
controlled by the site
Remote data
involves a different site
an account to access
requests require only data from one site
distributed database processing capabilities are not required
potentially useful for organizations operating
local control of computing resources
follows the geographical locations
contain additional components
nondistributed DBMSs
The local data managers
complete features of a DBMS
The distributed data manager
optimizes query execution
coordinates concurrency control
recovery across sites
controls access to remote data
the global dictionary to locate parts
can be homogeneous or heterogeneous
can call internal components
