|
|
Fig. 1: API & Operating System
|
 |
Software is needed for the computer to make use of the network's hardware. For example, the
Internet, a global network, utilizes software on the servers to send messages to your computer.
These messages, in turn, are captured by your network drivers and then handed over to your world
wide web browser, your mail program, or your multiplayer video game, which then outputs the
results on your computer.
Older systems used separate products such as Novell Netware, while today's operating systems,
like Windows NT and the Macintosh OS, integrate network support in its software. In the case of
the latter, the operating system works in conjunction with API's, or application programming
interfaces (API's).
Both of these enable applications to communicate with other
computers on the same transport layer. Applications are able to do this by invoking the API, which
does the actual sending and receiving of raw data. The API does not
address the packet with its destination and source, nor does it interpret the data to immediately
output it to the screen or convert it into another format (Burghart). The API simply passes on the message.
The operating system, on the other hand, takes the message and stamps it with information, such
as the source computer and network, the destination computer and network, the message's type,
the length, and error-control data. The message is then rerouted to the network communications
driver, the interface card driver, and then the actual interface hardware. (Derfler 87-88)
What seems to be merely passing messages is actually the lifeline of distributed computing. These
messages are crucial for distributed computing to work, whether you apply it to a terminal that
relies on its ties to the server or a home computer running SETI@Home, which needs to send back
results to the Internet server and take new, fresh data to process.
The preceding articles described a fundamental network, the communication of packets, and how
the packets are sent on a local level. In the next few sections, we will take a closer look at how
distributed computing works over local-area networks, and its derivatives — parallel processing
with a group of processors, and Internet distributed processing with the millions of processors.
Onward!
|