JargonFile/entries/timesharing.txt
2014-04-26 16:54:15 +01:00

39 lines
2.5 KiB
Plaintext

timesharing
Timesharing is the technique of scheduling a computer's time so that they
are shared across multiple tasks and multiple users, with each user having
the illusion that his or her computation is going on continuously. John
McCarthy, the inventor of LISP , first imagined this technique in the late
1950s. The first timesharing operating systems, BBN's Little Hospital and
CTSS , were deplayed in 1962-63. The early hacker culture of the 1960s and
1970s grew up around the first generation of relatively cheap timesharing
computers, notably the DEC 10, 11, and VAX lines. But these were only cheap
in a relative sense; though quite a bit less powerful than today's personal
computers, they had to be shared by dozens or even hundreds of people each.
The early hacker comunities nucleated around places where it was relatively
easy to get access to a timesharing account. Nowadays, communications
bandwidth is usually the most important constraint on what you can do with
your computer. Not so back then; timesharing machines were often loaded to
capacity, and it was not uncommon for everyone's work to grind to a halt
while the machine scheduler thrashed, trying to figure out what to do next.
Early hacker slang was replete with terms like cycle crunch and cycle
drought for describing the consequences of too few instructions-per-second
spread among too many users. As GLS has noted, this sort of problem
influenced the tendency of many hackers to work odd schedules. One reason
this is worth noting here is to make the point that the earliest hacker
communities were physical, not distributed via networks; they consisted of
hackers who shared a machine and therefore had to deal with many of the same
problems with respect to it. A system crash could idle dozens of eager
programmers, all sitting in the same terminal room and with little to do but
talk with each other until normal operation resumed. Timesharing moved from
being the luxury of a few large universities runing semi-experimental
operating systems to being more generally available about 1975-76. Hackers
in search of more cycles and more control over their programming environment
began to migrate off timesharing machines and onto what are now called
workstations around 1983. It took another ten years, the development of
powerful 32-bit personal micros, the Great Internet Explosion before the
migration was complete. It is no coincidence that the last stages of this
migration coincided with the development of the first open-source operating
systems.