Browse Prior Art Database

Project MAC: Excerpts from "Time-Sharing Measurement"24

IP.com Disclosure Number: IPCOM000129719D
Original Publication Date: 1992-Jun-30
Included in the Prior Art Database: 2005-Oct-06
Document File: 2 page(s) / 16K

Publishing Venue

Software Patent Institute

Related People

Allan Scherr: AUTHOR [+2]

Abstract

This article describes measurements made of timeshared system performances and user characteristics and discusses how they might be applied to the design of future systems. These measurements, taken on the Project MAC time-shared IBM 7094, were made as part of a larger study of analysis techniques for time-shared system performance.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 1 of 2

THIS DOCUMENT IS AN APPROXIMATE REPRESENTATION OF THE ORIGINAL.

Copyright ©; 1992 by the American Federation of Information Processing Societies, Inc. Used with permission.

Project MAC: Excerpts from "Time-Sharing Measurement"24

Allan Scherr

Allan Scherr, while involved as a graduate student with Herbert Teager on a parallel activity, executed extensive performance measurements on the first Project MAC system which indicate the service capabilities and usage characteristics of the transmuted C TSS system, then running on an IBM 7094. (Excerpts reprinted with permission from Datamation C) 1966.)

This article describes measurements made of timeshared system performances and user characteristics and discusses how they might be applied to the design of future systems. These measurements, taken on the Project MAC time-shared IBM 7094, were made as part of a larger study of analysis techniques for time-shared system performance.

The statistics presented are the results of measurements made during the three-month period from December, 1964, through February. 1965, and of simulations run thereafter.

Description of a MAC system

The MAC system, during the period of measurement, consisted of an IBM 7094 (Model I) with two 32K 2 msec memories. IBM 1301-2 discs, an IBM 7320A drum, and an IBM 7750 connected to Teletype Model 35 and IBM 1050 terminals. Other equipment is present (e.g., two tape channel is). but is not used during "normal" time-shared operation.

System saturation

The phenomenon of saturation can be defined in different ways. The author prefers to express saturation in terms of the utilization of the system. Specifically, as the probability that no interacting user is in Working or Command Wait state gets smaller, the system goes further into saturation. Since this probability may become infinitesimal but never absolutely zero, a system is never "completely" saturated. Thus saturation is relative, and saturation "point" must be defined accordingly.

The degree of saturation can be reduced in many ways: by limiting the total number of interacting users; by decreasing overhead and running times through more efficient system programming: by installing a faster CPU (or an additional processor); better swapping devices, etc.; by using slower consoles, etc.

(Image Omitted: Figure 1. Ratio of response time to processor time per interaction versus mean number of interacting users. (Reprinted with permission from Datamation C) 1966.))

Saturation can be clearly seen in the plot [Figure 1] of the ratio of response time to processor time per i...