Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Time Sharing Scheduling Method

IP.com Disclosure Number: IPCOM000074103D
Original Publication Date: 1971-Mar-01
Included in the Prior Art Database: 2005-Feb-23
Document File: 3 page(s) / 17K

Publishing Venue

IBM

Related People

Harrison, WH: AUTHOR [+2]

Abstract

In any system which multiplexes resources among more than one requestor, methods must be provided to resolve contention in a "fair" manner. Such methods are called "scheduling algorithms". In some time sharing programming systems (TSPS), major resources multiplexed by the system which have the greatest bearing on the rate at which a request is processed are: (1) use of the area of main storage to which the processing program has been assigned, and (2) the relative priority at which requests for central processing unit (CPU) and I/O resources are to be queued.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 38% of the total text.

Page 1 of 3

Time Sharing Scheduling Method

In any system which multiplexes resources among more than one requestor, methods must be provided to resolve contention in a "fair" manner. Such methods are called "scheduling algorithms". In some time sharing programming systems (TSPS), major resources multiplexed by the system which have the greatest bearing on the rate at which a request is processed are: (1) use of the area of main storage to which the processing program has been assigned, and
(2) the relative priority at which requests for central processing unit (CPU) and I/O resources are to be queued. Since the notion of "fair" in a time sharing system is unresolved, it is deemed necessary for the design of a TSPS to: (1) provide for the easy replacement of the scheduling algorithms, and (2) provide a highly parameterized set of algorithms which can be configured to fit most of the currently accepted notions of "fairness". TSPS Driver and its Inputs.

A TSPS driver program is provided as components in determining the "fairness" function in the TSPS. The driver receives as its input the signals indicating the occurrence of each of a large number of events. Whenever some component of the system detects or causes a condition which may be of relevance to the scheduling of the system, this condition is reported to the driver as an event. Several dozen such events exist within the following general classes:
1) Events associated with swapping,
2) Events describing which jobs currently have control of

the CPU,
3) Events relating to the ready/unready status of a job,
4) Events describing terminal interactions,
5) Miscellaneous other events.

The system signals an event to the driver by reporting the following information: the event number, which user (if any) the event is related to, the current time of day, and other information which depends upon the specific event in question. Outputs from the TSO Driver.

The outputs of the TSPS drive may signal any of three general classes of action:
1) Specify a new time at which the system should re-enter

the driver with a timer event,
2) Specify an alteration in the relative dispatching

priorities of time sharing users and of background jobs,
3) Specify that the user currently in some time sharing

region should be quiesced, or that some new user should

be swapped in. In addition, certain events elicit outputs specific to that event.

Time Sharing Algorithms.

The TSPS driver includes methods for assignment of three resources. These methods are termed: swap scheduling, CPU scheduling, and region selection.

Swap Scheduling: Since many users of a TSPS installation require use of the same set of addresses (called a "region"), and since an address may be

1

Page 2 of 3

assigned by the CPU to only one use at any given time, the TSPS driver treats each region as an allocatable resource whose use is multiplexed across many users of the system. Contention for this resource is represented by a series of queues.

In operation,...