--CLIENT-SERVER COMPUTING VERSUS TERMINAL MAINFRAME COMPUTING--
Back in the 1960s, 70s and early 80s, networks were run on mainframe computers. The clients were known as "dumb" terminals hooked up to mainframe computers, whose size can take up an entire room. Now, "dumb" terminals were just early versions of the PCs at the time, except the terminals totally relied on the mainframe for all processes-- the terminals didn't have a local hard disk or CPU like today's PCs. Then came the introduction of the Mac, DOS and eventually Windows PCs. (For the Mac Lovers-- yes, the Macintosh is considered a PC. PC stands for "Personal Computer", which a Mac is a category of. The acronym PC is usually associated with computers running DOS/Windows because when IBM made its foray into the Personal Computing Industry in the early 80s, they called it the "IBM PC", and that's how the label stuck.
In the late 80s and all of the 90s-- with the proliferation of Windows PCs-- the emphasize shifted to the Client-Server computing model. In this model, the client-- usually stocked with Windows as the OS and an Intel CPU-- performs some of the processes. The client still relies on a "server" for networked functions (file and printer sharing, etc.), but the client (a PC usually called a desktop or workstation) can also perform tasks on its own before calling on the network server.
The Client-Server model was a good idea because it allowed small and medium sized businesses to put together a network with purchasing proprieatary mainframe computers costing millions of dollars.