GridLab
Grid Application Toolkit

A simple API for Grid Applications
GAT

Menu



next up previous contents
Next: Peer-to-Peer Up: The Grid Computing Model Previous: The Grid Computing Model   Contents

Client-Server

What'd you feel? I felt the Grid as a ``client-server'' environment. The client, the application on which the application programmer toils, calls upon servers, usually large computing resources, to do its bidding, see figure [*]. A client-server application usually consists of a powerful server with CPU power and disk space to spare and a little client application that presents the results of the server's computational might to the end-user. In the vast majority of the cases, the client sends a request for the server to complete some computationally arduous task, such as finding the first billion digits in $\pi$, then waits for the server to send its result. Usually the request the client sends is for a computational task that the client could not complete with local resources; so, they have to ship the work out to foreign contractors.

The most popular client-server application by far is the World Wide Web. On the web a client, your browser, sends out a request, ``Send me such and such web page.'' The server responds by sending exactly that web page to your web browser which presents it to your waiting eyes. This may seem a simple example, but the production of some web pages may actually involve serious computation. For example, some high performance simulations carry within their bowls a little web server that is able to dish-out web pages. So, when a simulation is running, determining the weather in Japan in the year 2020 or whatever, you can, with a simple web browser, take a peek in on the progress of this simulation and determine if you should or should not by a rain coat in the year 2020.

Figure: Client-Server Model
[width=]ClientServer


next up previous contents
Next: Peer-to-Peer Up: The Grid Computing Model Previous: The Grid Computing Model   Contents
Andre Merzky 2004-05-13