Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!att!tut.cis.ohio-state.edu!cs.utexas.edu!uunet!mcsun!ukc!icdoc!syma!aarons From: aarons@syma.sussex.ac.uk (Aaron Sloman) Newsgroups: comp.arch Subject: Re: OS/2 is dead? (actually multiple logins) Summary: allowing extra logins can be very useful Keywords: OS/2, multi-user, networked Message-ID: <4103@syma.sussex.ac.uk> Date: 1 Jan 91 19:36:17 GMT References: <28775@usc> <14887@ogicse.ogi.edu> <5074@trantor.harris-atd.com> <1990Dec13.170613.19556@msuinfo.cl.msu.edu> Organization: School of Cognitive & Computing Sciences, Sussex Univ. UK Lines: 77 gould@pilot.njin.net (Brian Jay Gould) writes: > Date: 13 Dec 90 20:26:01 GMT > Organization: NJ InterCampus Network, New Brunswick, N.J. > > I can't understand why someone would want multi-user support on their desktop > workstation. I have frequently welcomed this on networked systems. (a) Every now and again the program I am interacting with gets into a state in which I can't do anything from the keyboard. By logging in "round the back" I can kill the offending process, and continue normally. Without this I'd have to re-boot, a slower procedure that would also kill other background jobs, that were running quite happily on my machine, etc. (b) My secretary, who needs only to access a text editor, read mail, give print commands, etc. can plug a VDU into the back of my machine, log in, and work happily without my even noticing any disturbance. (c) In a team where we are developing software on networked machines of different sorts, it is convenient for me to log in to my colleague's machine upstairs without leaving my desk, in order to check out how something runs on his machine and not mine. (d) When I log into our network via a terminal from home, it enables me to log into the machine on my desk in order to work from home. (e) One of my colleagues has a workstation that is quite a bit faster than mine. When he is not using it, I can log through to his machine and do some things faster. One of my students normally uses a slower machine than mine. I let him log into my machine when I am not using it, without having to give him the key to my office. (f) When I have a problem I can't sort out, I can sometimes phone or email a member of our systems support staff who then logs in to my machine via the network and fixes it, without having to leave his/her office. etc. etc. (Some of these points assume an integrated file-store accessible from several machines via NFS.) In my experience the people who can't imagine why they would want a multi-user machine on their desk are mostly people who have never had the good fortune to have one. However, I agree with the previous commentator (from Sweden?) who implied that for most multi-user networked environments the cost-effective way of the future (taking into account hardware costs, management costs, and the cost over several years of giving all users more and more speed and memory as the technology improves) is NOT going to be based on workstations, but on something like X terminals linked to very powerful, easily upgradeable, symmetric multi-processor CPU/memory/file servers. There will be a few kinds of users for whom workstations are a better solution, but only a minority in most organisations. However, the arguments in support of this view are somewhat complicated, and in my experience many people don't understand them because they are hung up with memories of time-sharing in the 1970s, or they complain that network traffic will be too high because they are hung up with problems of diskless workstations doing too much swapping, paging and context switching over their existing network! I suppose this is relevant to comp.arch only in the sense of being concerned with the architecture of a computing service? Aaron Sloman, School of Cognitive and Computing Sciences, Univ of Sussex, Brighton, BN1 9QH, England EMAIL aarons@cogs.sussex.ac.uk