Personal Distributed Computing: The Alto and Ethernet Hardware Charles P. Thacker Systems Research Center Digital Equipment Corp. 130 Lytton Avenue Palo Alto, CA 94301 Abstract Between 1972 and 1980, the first distributed personal computing system was built at the Xerox Palo Alto Research Center. The system was composed of a number of Alto workstations connected by an Ether- net local network. It also included servers that pro- vided centralized facilities. This paper describes the development of the hardware that was the basis of the system. Introduction In the last few years, a new type of computing en- vironment has become available. These distributed personal computing systems represent another step in the process, started by timesharing, of bringing computing power closer to the user. Although many variations are possible, these systems share a number of characteristics: • They are based on workstations-----personal ma- chines that are sufficiently powerful to satisfy es- sentially all the computational needs of a single user. The workstations include high resolution displays, and provide a highly interactive user interface. * The workstations are interconnected by local net- works that provide high bandwidth communica- tion throughout a limited area, typically a single building. • In addition to the workstations, the network contains serners---nodes that provide capabilities that need to be shared, either for economic or logical reasons. Timesharing systems grew primarily from the vi- sion of man-computer symbiosis presented by J.C.R. Licklider in a landmark 1960 paper [29]. Efforts to realize the possibilities presented in this paper occu- pied the creative talents of many computer science re- searchers through the sixties and beyond. Distributed personal computing systems build on this view of the way computers and people interact by providing a level of responsiveness that timesharing systems can- not achieve. The first distributed personal computing system was built at the Xerox Palo Alto Research Center over a period spanning 1972 to 1980. The workstation used in this system was the Alto [36]; the network was Ethernet. This paper describes the hardware that was the foundation of this system. A companion pa- per by Butler Lampson [24] describes the software that was built on the hardware base described here. This paper contains seven sections: Section 1 de- scribes the environment in which the work was done. Section 2 traces some of the underlying ideas. Section 3 describes the early implementation period, and sec- tion 4 discusses the servers that provide printing and file storage in the system. In section 5, the reengi- neering effort that made the Alto into a successful internal product is described. Section 6 briefly dis- cusses some of the Alto's successors, and section 7 contains concluding remarks. Permission to copy without fee all or part of this material is granted provided that the copies are not made or distributed for direct commercial advantage, the ACM copyright notice and the title of the publication and its date appear, and notice is given that copying is by permission of the Association for Computing Machinery. To copy otherwise, or to republish, requires a fee and/or specific permission. © 1986 ACM-0-89791-176-8-1/86-0087 $00.75 87 1 The Environment The Xerox Palo Alto Research Center (PARC) was established in 1970, primarily through the efforts of Jacob Goldman, director of corporate research. It was composed of three laboratories: the ComputerScience Laboratory, the System Sciences Laboratory, and the General Science Laboratory. To direct the new center, Goldman recruited George Pake, a physi- cist who was at that time provost of Washington Uni- versity, St. Louis. To establish the Computer Science Lab, Pake en- gaged Robert Taylor, who had directed the Informa- tion Processing Techniques Office of ARPA during the late sixties. Taylor had worked with and funded many of the leading computer science research groups during this period. As a result, he was in a unique position to attract a staff of the highest quality. During the first year of CSL, Taylor built a group of approximately fifteen researchers, drawn from IV[IT, the University of Utah, and CMU. Several members of CSL came from Berkeley Computer Corporation, a start-up company composed primarily of individ- uals from the University of California at Berkeley, who had built one of the first timesharing systems [25]. From Bolt, Beranek and Newman, Taylor later brought Jerome Elkind, who was the manager of CSL from 1971 until 1976. Also during this period, Alan Kay, who was to provide much of the vision on which the Alto was based, was recruited into the Systems Sciences Laboratory by Taylor. Kay estab- lished the Learning Research Group (LRG), and de. fined its goal: To produce a programming system in which 4... simple things would be simple, and com- plex things would be possible [20]." The research environment built by Taylor was one of the main reasons for the success of CSL and its projects over the next decade. Unlike other PARe laboratories, CSL was not organized into permanent groups. Instead, researchers were encouraged to move between projects as their talents and the needs of the projects dictated. This flat structure and the mobil ity it made possible encouraged members of the lab to become familiar with all activities. Additionally, it provided a continuous form of peer review. Projects which were exciting and challenging obtained some- thing much more important than financial or adminis- trative support; they received help and participation from other CSL researchers. As a result, quality work flourished, less interesting work tended to wither. During 1971 and early 1972, most of the effort in CSL was spent in building a set of hardware and soft- ware facilities to support the future work of the lab- oratory. The IV[axc timesharing system [13] was built and the Tenex [2] operating system was acquired and modified for it. Projects in graphics, computer net- working, and language design were started.
View Full Document
Unlocking...