Browse Prior Art Database

A POSTMORTEM FOR A TIME SHARING SYSTEM

IP.com Disclosure Number: IPCOM000128898D
Original Publication Date: 1974-Jan-01
Included in the Prior Art Database: 2005-Sep-20

Publishing Venue

Software Patent Institute

Related People

Sturgis, H.E.: AUTHOR [+3]

Abstract

In 1968 the University of California Berkeley Campus Computer Center began a project to design and implement a Time Sharing System for a Control Data Corporation (CDC) 6400 computer, with Extended Core Store (ECS). The project continued until the Fall of 1971 when It was terminated due to a lack of funds. The author WAS a member of the project from the beginning, and was director at its termination. The system we designed, CAL TSS, included a number of ideas proposed by other projects, that had not yet been fully testes. These included the concept of capability based protection (system maintained pointers to system objects, through which all access to those system objects must pass), and a mapped address space (all storage resides in files, and all load and store machine instructions actually access data in some file, rather than in a local memory). In contrast to other projects, such as Multics [ C3 ] ], this was a small project. At its peak, there were about eleven people involved, many part time. This thesis contains a discussion of some of our underlying ideas, describes the system we constructed and finally some reactions to that system. Part One describes the project, the hardware and the underlying ideas. Part Two describes the system. Finally, Part Three contains my reactions to various aspects of the system.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 4% of the total text.

Page 1 of 38

THIS DOCUMENT IS AN APPROXIMATE REPRESENTATION OF THE ORIGINAL.

©; Xerox Palo Alto Research Center, 1974

A POSTMORTEM FOR A TIME SHARING SYSTEM

BY HOWARD EWING STURGIS

CSL 74-1 JANUARY, 1974

This thesis describes a time sharing system constructed by a project at the University of California, Berkeley Campus, Computer Center. The project was of modest size, consuming about 30 man years The resulting system was used by a number of programmers. The system was designed for a commercially available computer, the Control Data 6400 with extended core store. The system design was based on several fundamental ideas, including:

specification of the entire system as an abstract machine, a capability based protection system, mapped address space, and layered implementation.

The abstract machine defined by the first implementation layer provided 8 types of abstractly defined objects and about 100 actions to manipulate them. Subsequent layers provided a few very complicated additional types. Many of the fundamental ideas served us well, particularly the concept that the system defines an abstract machine, and capability based protection. However, the attempt to provide a mapped address space using unsuitable hardware was a disaster. This thesis includes software and hardware proposals to increase the efficiency of representing an abstract machine and providing capability based protection. Also included is a description of a crash recovery consistency problem for files which reside in several levels of storage, together with a solution that we used.

XEROX PALO ALTO RESEARCH CENTER 3180 PORTER DRIVE/PALO ALTO/CALIFORNIA 94304

ACKNOWLEDGENENTS1

First, I thank Professor James Norris, my dissertation committee chairman, for many hours of discussions and painstaking reading of many drafts. Second, I thank the other members of my dissertation committee, Professor R. S. Fabry and Professor Hartin Graham. Also, l thank all of the others who have read early drafts and commented extensively, including Dr. Butler Lampson, David Redell, Br. James Gray and Paul McJones.

For typing many drafts I thank Janet Farness, and for the illustrations Carl Stewart and Jackie Southern. The Xerox Corporation has Given me generous support while writing this dissertation.

A generous thanks is due to a faculty member who rekindled my interests, at a time when I had become discouraged with the university, Professor R. Sherman Lehman.

Finally, I thank my wife, Susan Sturgis, for her years of patience while I followed the rather tortuous path that led to this dissertation.

Howard Sturgis

1 While writing this dissertation, the author was employed by Xerox Corporation, Palo Alto Research Center, Palo Alto, California. The project was supported, in part, by a National Science Foundation Grant, UP 7635.

Xerox Corporation Page 1 Jan 01, 1974

Page 2 of 38

A POSTMORTEM FOR A TIME SHARING SYSTEM

Woodside California

Hay 1973

TABLE OF CONTENTS I

PART ONE: ASSORTED INITIAL CONSIDERATION...