Network Distributed Processing


The structure of the distributed processing network can significantly
affect its functionality, cost reliability, and performance. The first
distributed processing architecture using network data communications
was the file server approach. Its limitations gave rise to
client/server approach. This is the structure widely used by businesses to downsize from centralized mainframe computer processing. The third common approach is peer-to-peer. (Turban, McLean and Wetherbe, 1996)


File Server Architecture

File server distributed processing is the simplest approach for a relatively large number of network processors to share data. Enterprise-wide data are digitally stored in a designated file server processor connected to the network. When an application at any node needs to perform processing on data contained in a file, the application requests the file from the file server. Then, the file server responds by transmitting the entire file to the requesting node processor, and the requesting node performs the necessary processing locally. There are few limitation of this arrangement which helped to spur the development of the client/server computing architecture.

First, large files can electronically choke the network whenever they are transmitted to a node for processing, especially if they are transmitted often to a large number of nodes.
Second, many nodes can be simultaneously altering the same file which consequently create chaos when the file is "returned" to the file server.
Thirdly, the file server can only allow one node at a time to update a file.


Client/Server Architecture

Client/Server(C/S) refers to computing technologies in which the hardware and software components, i.e., clients and servers are distributed across a network. This technology includes both the traditional database-oriented client/server technology, as well as more recent general distributed computing technology. Besides, a client/server system is a user-centric system that emphasizes the user's interaction with the data. Client/server computing splits processing between "clients" and "servers". The users will experience the network as a single system with all functions, both client and server, integrated and accessible.

The client is the user point-of-entry for the required function in a client/server computing application. Normally a desktop computer, workstation, or laptop computer. The user generally interacts directly with only the client portion of the application, typically through a graphical user interface. The user typically utilizes it to input data and query a database to retrieve data. Once the data have been retrieved, the user can analyze and report on them, using fourth-generation packages such as spreadsheets, word processors, and graphics applications available on the client machine on the user's own desktop.

The server satisfies some or all the user's request for data or functionality and might be anything from a supercomputer or mainframe to another desktop computer. Servers store and process shared data and also perform back-end functions not visible to users, such as managing peripheral devices and controlling access to shared databases.

Every organization has its own data handling and data processing requirements. For instance, data communication at Northern Bank requires sharing data between three branch offices, located in two adjacent towns, and a credit card authorization center. Automatic teller machines (ATMs) and workstations are hardwired with coaxial cable to a server in each branch. Northern Bank uses a flexible client/server network design. In a client/server network, the user's computer, the client, takes on more responsibility than it does in a traditional server-oriented network. The client computer handles the user interface software. Northern's ATMs and personal computer is a diskless workstation and does not have disk drives because the data is maintained by the server.

The advantages of client/server computing include user convenience, scalability, and greater ability to accommodate and maintain hardware and software from different vendors.

Further more, there is another category of tool between the client tools and server tools called middleware. Middleware controls communication between clients and servers and performs whatever translation is necessary to make a client's request understandable to a server device. It provides required services such as remote database access, interprocess communication, distributed object management, directory services, and security services. Products and technologies in this area include:

The Common Object Request Broker Architecture (CORBA) from the Object Management Group (OMG),
The Distributed Computing Environment (DCE) from the Open Software Foundation (OSF),
The Component Object Model (COM) and OLE 2.0 from Microsoft, and
Microsoft Open Data Base Connectivity (ODBC)


Peer-To-Peer Architecture

Peer-to-peer architecture is an important alternative to client-server for small computer networks. In peer-to-peer, each workstation can communicate directly with each other workstation on the network without going through a specialized server. Peer-to-peer is appropriate when the network users mostly do their own work but occasionally need to exchange data. In these cases, it may be more efficient to keep data and copies of the software at each workstation to avoid the delays of downloading data and software each time a user gets started. However, peer-to-peer also has potential problems in security and consistency. For example, with data at someone else's workstation, the data may be difficult to retrieve when that person is out of the office and the workstation is shut off.

Home