Aucbvax.5839 fa.works utcsrgv!utzoo!decvax!ucbvax!works Mon Jan 18 15:06:13 1982 WorkS Digest V2 #6 >From JSOL@USC-ECLB Fri Jan 15 03:30:44 1982 Works Digest Friday, 15 Jan 1982 Volume 2 : Issue 6 Today's Topics: RFC - Book Layout Food For Thought - WorkStations Query Query Replies (a few of them) - What Is A WorkStation Laser Printer - Based On The SUN Board ---------------------------------------------------------------------- Date: 14 Jan 1982 (Thursday) 1019-EDT From: DREIFU at WHARTON-10 (Henry Dreifus) Subject: Possible layout for Workstations book: comments? PERSONAL COMPUTER WORKSTATIONS Chapter 1: What is a Personal Workstation - General Introduction - Its impact - List the components, brief description Chapter 2: Management & Economics of PWS (Maybe 2 chapters) - Why there is a trend towards PWS's - Where they belong, their function - What they should NOT be used for - Advantages, incremental growth, dist'd topologically - Costs, what they are worth, how much should one pay for a PWS - What makes a WS a justifiable expenditure - Businesses of the future - New management styles - Other changes due to PWS's Chapter 3: Expectations - Hardware, limited capabilities; it is not a mainframe - Software, it will be much better - What a PWS will and will not do - Misconceptions Chapter 4: [POSSIBLE] Design of a PWS - What does one need in a PWS - What NOT to put in a PWS - Expansion considerations - Reliability of WS, needs for new kinds of procedures for PWS's (archiving, backup, security, location) Chapter 5: Distributed Systems & Local Networks [could be 2 chapters] - Definition - How they work - Architecture - BroadBand, Baseband, what can one put on a BBN - What constitutes a distributed envr. Why does one need one - Problems, management & other babble w/dist sys. (concurrency) Chapter 6: Connecting up a PWS - Interface considerations, problems - Lack of Standards, types of communications, modes, bandwidth - Gateways - What does one attach to a PWS? How to set up a PWS in an envr, some example(s). - +'s and -'s of networks, what kinds: Broad v. Base band - Other devices, Laser Beam Printers, special devices Chapter 7: The User Interface - Display Managers - Personalization - User-oriented devices, representations. - Heterogeneous environments - User environments ] - Programmer environments ] their needs Chapter 8: The Workstation's Next Step - Now that we've got them, what else can they do for us - Training considerations - How do we "use" the workstation - Office of the Future - Engineering for the future (throw away drafting introduce CAD) - Impact of the WS DESCRIPTION PERSONAL COMPUTER WORKSTATIONS, tentative title, will be a book aimed at the technical and management market. Introducing Personal Workstations, defining some of the terms, identifying problems that exist, the goal is a leading edge informative book, from which a manager will understand what personal workstations can accomplish and how they can best be utilized in his environment. ------------------------------ Date: Thursday, 14 January 1982 11:06-EST From: DPR at MIT-XX Subject: questions I'd like to answer Generally, I find myself bored with the content of WorkS. THus, to spice it up let me ask some questions that I care about. 1. Rather than worrying about chips I care about the packaging of those chips into workstations. What are the best 68000 based products people have seen out now that so many are being announced (Computex, Dual, Wicat, Charles River Data Systems, BBN's BitGraph...)? 2. Is a multi-processor workstation a good idea (for dynamic graphics, for example)? Adding additional hardware processors with shared memory doesn't seem to add much to hardware cost, so why has no one done it? 3. I hypothesize that sound output on a workstation is cute, but a terrible product idea (in shared offices, it is annoying...). For similar reasons sound input is a problem. Is there a good use for sound in an office? 4. How do you train users of winchester-based systems to do proper backup so that they don't lose everything the first time their disk fails? Is there a technical solution (require a local net with a backup server?) 5. There seem to be two distinct design points for workstations regarding theeir communications environment -- either they are always connected to a net and always listening (for mail, e.g.), or they are only connected when the user is running a communications appl. (such as FTP or fetching mail from a mail drop server). Each is workable, but system usability, hardware design requirements (e.g., to always be listeening requires a kind of "personal timesharing system"), network technology (ehternet vs. digital pabx) are all affected by the choice of design point. Can anyone comment on their experience with one choice or the other; is one clearly better? is one cheaper? David ------------------------------ Date: 13 Jan 1982 00:10:42-PST From: pratt@Shasta at Sumex-Aim Subject: asc answered (again) To answer yet another asc question: What is a workstation? I haven't a clue, making me suspect it is the wrong question. So I'll change the question to "Where should computing resources be located, and at what time granularity and community size should they be shared?" Without going into great detail, here's an institutional answer based on today's prices and performance figures. Processors and main memory should go on the user's desk, in a corner of the box that the display goes in. Secondary memory should consist of perhaps 150 megabytes shared by 5-10 users, connected say by one Ethernet. Tertiary memory is harder to pin down, but might consist of a few gigabytes shared by 100 users. Printers (which should of course be laser printers) should be shared by one floor's worth of people - it is a pain to have to retrieve printouts from another floor. This is the model on which the Sun workstation design has been based. The design gets its economical leverage (a) by respecting this model, and (b) by parsimonious implementation. For noninstitutional users effective sharing is much more awkward to arrange, putting at a disadvantage anyone who needs to use a computer at home, whether or not they also have access to a computer at work. Either much cheaper memory, or much higher bandwidth communication with shared resources, will alleviate this problem. Of the two, better communications is the better solution, one reason being that software you can really benefit from is not the sort of thing you generally want to be maintaining on your own, even if you have a 100 megabyte disk to maintain it on. The above pretty much circumscribes "the system." If you can do better than this you are either a hobbyist (no slight intended, I am both a computer hobbyist and a ham - VK2AUA - myself) or are willing to settle for less than productive computing resources. Don't forget to figure in the value of your time when calculating the cost of lost productivity. Also don't forget that "usable" computers and memories are getting pretty cheap - we hope to get the Sun workstation design down to $1500 worth of parts including packaging within the next year or two, to bring the retail cost down to well under $5000. Vaughan Pratt ------------------------------ Date: 12 Jan 1982 11:55:34-PST From: decvax!yale-comix!ima!johnl at Berkeley From: John R. Levine From: The INTERACTIVE Electric Calculator Co., Cambridge MA. Subject: SUN workstation Friends at Yale just got a nifty laser printer that is based on a SUN board. Contrary to earlier claims, no paging. Perhaps they're just waiting for the 68010 and 68020. ------------------------------ Date: 14 January 1982 01:23-EST From: Robert Elton Maas Subject: WorkStations? To: CROM at MIT-AI 90% of what a programmer does at his terminal is editing (modifying and viewing) text (source programs, documentation of programs, expository articles about research, electronic mail). Thus a workstation should be defined as something that can do this 90% totally locally and call up a network for the rest. 100% of what a secretary or publisher does is editing text (financial data, correspondence, articles). Thus a workstation should be defined as something that can do all this and then send the result out to the postal service (correspondence) or the printing press (articles). ------------------------------ Date: 14 January 1982 05:25-EST From: Frank J. Wancho Subject: What's a WorkStation Seems to me that the definition is a function of the kind of work that needs to be done. For the most part, the recent discussion centers around S&E applications programming as the work - and there are those who will say, won't a relatively dumb terminal connected to a super mainframe do the job... But, there are other kinds of work stations that are not being discussed here, and maybe should be. One is the manager's work station. He's the one who wants to query locally shared databases (through a local net) as well as some piece of much larger databases elsewhere. He also wants to draw plots/graphs of the extracted data, both on the screen and occassional hardcopy, maybe at times, in color, for a report or presentation. He also wants to write draft reports, papers, and corespondence to have his secretary cleanup (format) and print - and maybe even send to someone else (via electronic means). Another is the office manager's work station. Mostly what the office secretary will use to prepare "paper" work and correspondence for inter-office and inter- and intra-site delivery... the first step toward the "paperless" office. And the last one of interest (to my people) is the project engineer's work station. His is the combination of all of the above, although the programming requirements are more routine, such as massaging data to prepare technical reports. Like the manager, he also needs to access the large databases to prepare his reports too... And all these somewhat different types of work stations need to communicate with each other and the outside world - and for a total cost of less than $5K per user - not $10-$30K... And do all of that today (with whatever's currently available, as an interim off-the-self solution that is not self-obsolescent), maybe tomorrow (*if* there is something really better and cheaper worth waiting for). Now, we have some ideas along those lines and we need to know if we've overlooked somethings in either the choice of the hardware or the choice of the software (operating system). Given that most of the WorkS discussions have been rather highly technical and leaning heavily toward the kind of work station that supports mainly heavy S&E applications programming, am I alone and maybe in the wrong discussion group? --Frank ------------------------------ Date: 14 January 1982 2049-EST (Thursday) From: George.Coulouris at CMU-10A Subject: What is a workstation? Given that we want personal machines that offer a natural and convenient user interface for all of the information handling tasks that we and other, possibly more naive, users want to do, I believe that we can derive some important consequences. A natural user interface to any reasonably sophisticated task must contain a visual representation of the state of the task as it progresses. If the task is complex, the visual representation needs to be quite rich in order to represent its state. It is for this reason that high-resolution displays have become so popular, not because all workstations are used for typesetting, vlsi design, or any other specific interactive graphical purpose. The visual representation needs to change in real time, so that it constitutes a 'window onto the state of the application'. For many applications, this involves animation-like image generation, *directly from the application data structures*. It therefore follows that the application program that manages the application data structures should run in the workstation, so that the data structures are available to the 'animation process'. I am therefore led to the conclusion that a workstation must have sufficient resources to execute a very large proportion of the interactive tasks its users wish to perform, leaving only non-interactive and very infrequently used software to run in other kinds of computer system. Of course, the workstation programs may call upon other stations to perform services (shared file access, printing, mailing, etc) but the state of the task has to stored in the workstation if a smoothly integrated interactive environment is to be achieved. This leads me to the further conclusion that workstations need as many of the architectural features that have been found really useful in current machines as we can afford to put into them, so that they can run the application software well. In addition, they need extra hardware support for the screen in order to achieve the above mentioned smooth animation. George Coulouris (Computer Systems Laboratory, Queen Mary College, London) ------------------------------ End of WorkS Digest ******************* ------- ----------------------------------------------------------------- gopher://quux.org/ conversion by John Goerzen of http://communication.ucsd.edu/A-News/ This Usenet Oldnews Archive article may be copied and distributed freely, provided: 1. There is no money collected for the text(s) of the articles. 2. The following notice remains appended to each copy: The Usenet Oldnews Archive: Compilation Copyright (C) 1981, 1996 Bruce Jones, Henry Spencer, David Wiseman.