Time-sharing

Time-sharing
Unknown authorUnknown author · Public domain · source
Name	Time-sharing
Caption	Early interactive terminals and mainframe cabinets
Introduced	1960s
Designers	John McCarthy (computer scientist), Fernando J. Corbató, Tom Kilburn, D. M. Ritchie
Platforms	CTSS, MULTICS, Compatible Time-Sharing System
Influenced	Unix, TENEX, Multics

Time-sharing is a technique for sharing computational resources among multiple users and processes by multiplexing processor time, memory, and I/O so that interactive tasks appear simultaneous. Pioneered in the 1960s, time-sharing transformed access to large mainframe computers, enabled interactive computing for researchers and students, and influenced the design of modern operating systems and networked services. It bridged batch processing eras exemplified by systems at IBM and interactive eras represented by projects at MIT, Bell Labs, and AT&T laboratories.

History

Early work on interactive, multi-user operation occurred at institutions such as MIT, Bell Labs, University of Cambridge Computer Laboratory, and Stanford University. Notable milestones include the Compatible Time-Sharing System (CTSS) at MIT Project MAC led by Fernando J. Corbató, the design of MULTICS by a consortium of MIT, General Electric, and Bell Labs, and experiments at Manchester University by Tom Kilburn. Influential figures included John McCarthy (computer scientist), who advocated for time-sharing and batch processing alternatives, and Douglas Engelbart whose interactive demonstrations leveraged shared computing. Commercial and government adoption accelerated with systems from IBM (such as CP/CMS research) and implementations at National Physical Laboratory (United Kingdom), while academic collaborations fostered standards and publications through venues like ACM and IEEE conferences.

Architecture and Concepts

Core architectural elements include a preemptive scheduler driven by a system clock, privileged kernel modes in operating systems, hardware support for context switching such as interrupt controllers and memory management units exemplified in designs by Intel and DEC, and virtual memory abstractions pioneered by Frederick P. Brooks Jr.'s contemporaries. Concepts central to time-sharing are process isolation via address translation tables, protection rings similar to those in Intel 80286 and later x86 families, terminal-oriented I/O via devices like Teletype machines and CRTs, and the use of swap space influenced by research at Bell Labs and Project MAC. System calls and user/kernel boundaries defined in UNIX and Multics reflect time-sharing heritage, while security models and access control trace to early implementations at institutions such as RAND Corporation and SRI International.

Scheduling and Resource Allocation

Schedulers mediate CPU allocation using algorithms ranging from round-robin, shortest remaining time, multi-level feedback queues developed in research at Carnegie Mellon University, to priority systems used in DEC VMS. Memory allocation employs segmentation and paging strategies influenced by Harvard University and Princeton University research. I/O scheduling and device arbitration originated in designs from IBM mainframes and were refined in academic projects at Stanford Research Institute. Quotas, fair-share scheduling, and admission control emerged from deployments at NASA centers and corporate data centers run by AT&T and Bell Labs to manage concurrent user loads and batch-job interactions. Real-time extensions and soft real-time guarantees later drew on work at MIT Lincoln Laboratory and industrial labs like Honeywell.

Implementations and Systems

Prominent implementations include Compatible Time-Sharing System (CTSS), MULTICS, TENEX at BBN Technologies, and early UNIX variants developed at Bell Labs by Ken Thompson and Dennis Ritchie. Commercial time-sharing services were offered by companies such as Service Bureau Corporation and General Electric's computer division. Minicomputer-era systems from Digital Equipment Corporation like RSTS and VMS provided multi-user capabilities, while academic systems such as SDS Sigma series and installations at Lawrence Livermore National Laboratory demonstrated scaling strategies. Later virtualization platforms such as VMware and Xen inherit scheduling and isolation concepts from classical time-sharing.

Applications and Impact

Time-sharing enabled widespread interactive use in research labs, university campuses, and corporate settings, catalyzing developments in software engineering, human–computer interaction exemplified by Douglas Engelbart's work, and programming languages like BASIC and FORTRAN adapted for interactive use. It supported collaborative projects at ARPANET nodes and influenced the rise of networked services such as online service bureaus and early email systems at BBN Technologies and SRI International. Educational computing programs at institutions such as Stanford University and MIT expanded access to computation, while commercial services influenced regulatory and business practices at firms like IBM and AT&T.

Criticisms and Limitations

Critiques of time-sharing included concerns about security and isolation highlighted by incidents investigated by RAND Corporation analysts and National Bureau of Standards researchers, performance overheads from context switching noted in evaluations at IBM and Bell Labs, and the difficulty of providing strong guarantees for real-time tasks as raised in studies at MIT Lincoln Laboratory. Economic critiques focused on the cost structures of early service bureaus like Service Bureau Corporation versus dedicated personal systems promoted by companies such as Apple Computer and Microsoft. Scalability limits and the transition to distributed computing models challenged centralized time-sharing assumptions in projects at Xerox PARC and later in cloud computing efforts by Amazon Web Services and Google.

Category:Operating systems