Alan Turing

Architecture client/server indicates means of communication between several Ordinateur S of a network which distinguishes one or more stations customers from the waiter: each Logiciel customer can send requests to a waiter. A waiter can be specialized in Serveur of applications, files, terminals, or of email.

Characteristics of a waiter:

  • it is passive (or slave);
  • it is with listening, ready to answer the requests sent by customers;
  • as soon as a request reaches him, it treats it and sends an answer.

Characteristics of a customer:

  • it is active (or Master);
  • it sends requests to the waiter;
  • it awaits and receives the answers of the waiter.

The customer and the waiter must of course use same the Communication protocol. A waiter is generally able to serve several customers simultaneously.

Another type of architecture network is the Poste at station (or peer-to-peer in English), in which each computer or software is at the same time customer and waiter.

Architectures three thirds and multi-third

The terms “three thirds” and “multi-third” are wrongly translate of English three tier and multi-tier or n-tier .

Architecture client/server has two types of computers on a network: the customers and the waiters, it thus has two levels and is called two-tier in English. Architectures multi-third (or distributed) divide the waiter into several entities (for example, a waiter of application which itself is the customer of a database server).

Advantages compared to distributed architectures

  • All the data are centralized on only one waiter, which simplifies the security checks and the update of the data and the software;
  • technologies supporting architecture client/server are more mature than the others.

Disadvantages

  • Si too many customers wants to communicate with the waiter at the same time, this last risk not to support the load (whereas networks P2P go better while adding again participating);
  • if the waiter is not available any more, more none the customers goes (network P2P continues to go, even if several participants leave the network).

Examples

  • the consultation of pages on an Internet site functions on an architecture client/server. A Net surfer connected to the network via his computer and a Navigateur Web are the customer, the waiter is consisted the computers containing the applications which deliver the required pages. In this case, it is the communication protocol HTTP which is used.

  • the emails are sent and received by customers and are managed by a mail server. The protocols used are smtp, and the POP or IMAP.

  • the management of a database centralized on a waiter can be done starting from several stations customers which make it possible to visualize and seize data.

  • the system X Window functions on an architecture client/server. In general the customer turns on the same machine that the waiter but can be as well launched on another computer belonging to the network.

See too

  • Architecture three thirds

Structure three third thirds and N

Architecture client/server is current in the applications where many users must reach same a Database. , In these applications, several software layers are distinguished:

  • the layer presentation (i.e. customer) which is the part of the software that one finds on the side of the end user. As its name indicates it, this part contains only the user interface (generally, an graphical interface under X Window, ms Windows or Mac OS)
  • the layer of management of the data (the waiter), which consists of a Database or any storage system of the data
  • the logical layer which is the heart even of the application; it establishes the link between the customer and the waiter, and forms what one calls the Serveur of application.
This type of architecture in three layers is called Architecture three thirds .

However, the logical layer is not necessarily insulated: indeed, it can be very well moved towards the waiter (in form, for example, of procedures stored in the database), or towards the customer (all the application being then on the station customer, one does not make any more but require data of the waiter, via for example ODBC or SQLNet…). Such an architecture in two layers is called 2-third .

Architectures N-third are of course possible, by combining one or more waiters of application (offering each distinct service) and one or more databases.

The physical site of the various layers composing such an application is completely variable.
In a 3-third application, for example, the three layers can be:

  • three distinct programs functioning on the same machine
  • the customer on a machine, the two other layers on a second machine
  • three programs functioning each one on a distinct machine.

The theoretical advantage of the customer-server applications is the possibility of replacing a layer independently of the others: to be able to migrate for example a customer part of Visual BASIC towards PowerBuilder without having to modify neither the database, nor the waiter of application. But in practice, all is not so pink: the layers are actually seldom completely independent from/to each other, and the modification of a layer very often involves modifications in the others. In addition, the complexity of the application proportionally increases with the number of layers to develop and make reliable. Lastly, the maintenance of such applications is more complex:

  • when an error occurs, it is not always obvious to determine the faulty layer.
  • when a new version of the application must be delivered to the users, all the layers must be delivered simultaneously to avoid the problems of desynchronization between the versions of the customers, the waiter (S) of application and of the waiter (S) of data.
-->

Random links:Photon mapping | FK Moscow | Rotherham R.U.F.C. | Amphitheater of Catane | Vassili Barsoukov