🚀 go-pugleaf

In order to define client-server computing we have to be practical
about it.  It doesn't exist just for it's own sake, it accomplishes or
enables something to happen. I believe client-server is based on moving
processing to processors that can do it the best. By the best, I mean
at the least cost for processing and network. This moving of processing
can be called partitioning. An application can be partitioned to
optimize it's performance. The pieces we partition are conventionally:

o Presentation
o Processing
o Data

Presentation is the display portion, everything necessary to draw the
screen and present the data. This is the part the user sees. Processing
is the unseen part of computing. The manipulation that an application
must do to move, compute, and prepare data for Presentation or Data.
Data is the data management piece, mostly represented by Data Base
Management Systems, but also could include simpler file management.

Using these ideas, that we want to partition the three pieces of
computing to optimize an application, we can see how a simple example
of client-server evolved. Look at the ATM machines. They obviously are
mostly Presentation. They can't all be expected to have the Data piece,
so they are probably partitioned (simplified) as follows:

Client ATM's
	Presentation
	Processing <---+
                   |
Bank Server(s)     |
	Processing <---+
	Data

The Client and Server are then communicating their transaction at a
Processing to Processing level, using whatever protocols or method is
available and useful for the network. The important part is that the
servers do not have to worry about Presentation protocols on the
ATM's.

There is clearly a lot more to client-server, but the important concept
is partitioning of Processing, Presentation, and Data.

Stephen W. Hendershott
AT&T
steveh@devildog.att.com

-- 
Stephen W. Hendershott
steveh@devildog.att.com