[comp.arch] Needs of Clerical Users.

davecb@yunexus.UUCP (David Collier-Brown) (10/24/89)

In article <76700077@p.cs.uiuc.edu>, gillies@p.cs.uiuc.edu writes:
|  But most office clerical work needs no more than an XT 8088 machine
|  and WordPerfect.  They don't need color, SPICE, or a large database.
|  Let's hope these clericals are doomed, or the PC industry will die as
|  they flourish.

davidsen@crdos1.crd.ge.COM (Wm E Davidsen Jr) writes:
|  Sorry, you're making poor use of your clerical workers. SPICE they
| don't need, but access to databases, color highlighting for better
| productivity, and bit mapped graphics to display pictures and drawings
| all boost productivity.

  I can give a good example of nominally "clerical" workers needing more
capacity than a low-end pc...

  At Ork University, we provide clerical staff and first-level supervisory
staff with either PCs attached to a ring or terminals attached to VAXen.
The PCs have various office-productivity tools.  The VAXen have mail.
Both of them bottleneck their users.

  This is most obvious when you are discussing the first-level supervisors.
We use MMIPS as a measure of required processor power: MIllions of Mail
messages Per Second. (See, this does have something to do with architecture).
They need exactly the same facilities I use to manage 8 timesharing machines.
They need status indicators for mail to various mailboxes (projects,
machines), multiple windows so they can can work on more than one message
or paper at any one time, and the kind of clarity and display quality that
you get with an older workstation, not a PC.  My supervisor needs either
an X-terminal or my Sun 3/50.  

  The junior clerical staff also need the ability to work on more than
one thing at a time, or they have to resort to scribbled notes about
what other things they can't work on because we've shortchanged them.

  All in all, the low-end users need the same **kinds** of things everyone
else does.  We elect not to spend much money on them because we can't
always recoup it in staff reductions (the non-productivity of knowledge
workers is the subject of postings elsewhere).

  Architecturally, the low-end worker needs an inexpensive processor with
moderate power, a large (non-convoluted portion of an?) address space,
normal interprocess protections and software to run on the machine.
This is a niche which small manufacturers can play in (eg, low-cost new
machines compatible with older, slower workstations), and manufacturers
can supply (eg, almost-fast enough chipsets, well-amortized older
production products).  It does require that the components evolve to
track their high-cost bretheren (so one can use this years software on
them: 68010s need not apply!).

--dave
-- 
David Collier-Brown,  | davecb@yunexus, ...!yunexus!davecb or
72 Abitibi Ave.,      | {toronto area...}lethe!dave 
Willowdale, Ontario,  | Joyce C-B:
CANADA. 416-223-8968  |    He's so smart he's dumb.

henry@utzoo.uucp (Henry Spencer) (10/26/89)

In article <4576@yunexus.UUCP> davecb@yunexus.UUCP (David Collier-Brown) writes:
>  Architecturally, the low-end worker needs an inexpensive processor with
>moderate power...  It does require that the components evolve to
>track their high-cost bretheren (so one can use this years software on
>them: 68010s need not apply!).

Just to be difficult :-), I would contend that this is exactly wrong.
It will not be economical to provide processors to the low-end worker
unless one can be confident that they will *not* need to track the headlong
progress of high-cost computers... because such tracking is expensive, too
expensive for the bulk of the low-end-worker market.  For that market, one
needs to be able to make an investment that will be good for 5-10 years.
Always needing to have the latest thing on your desk makes Sun very rich
and you very poor.

I think Rob Pike has the right idea:  put something positively doddering,
like a 68010 :-), on the desks, and have it do *nothing* but interaction.
Centralize the heavy computing where it can be updated centrally in a
relatively economical way.
-- 
A bit of tolerance is worth a  |     Henry Spencer at U of Toronto Zoology
megabyte of flaming.           | uunet!attcan!utzoo!henry henry@zoo.toronto.edu

davecb@yunexus.UUCP (David Collier-Brown) (10/26/89)

This is now a discussion of tracking changes in architecture: should it be
moved to a different forum?  --dave


| In article <4576@yunexus.UUCP> davecb@yunexus.UUCP (David Collier-Brown) writes:
|   Architecturally, the low-end worker needs an inexpensive processor with
| moderate power...  It does require that the components evolve to
| track their high-cost brethren (so one can use this years software on
| them: 68010s need not apply!).

henry@utzoo.uucp (Henry Spencer) writes:
| Just to be difficult :-), I would contend that this is exactly wrong.
| It will not be economical to provide processors to the low-end worker
| unless one can be confident that they will *not* need to track the headlong
| progress of high-cost computers...

  Hmmn...
  This raises the question of which model of computing is best suited
to this "class" of user.  Distributed storage & processing is good for the
autonomous(sp?) user, centralized for the "pool".  Lets not...

  Restricting ourselves to people who will usefully use an older workstation
forces us to track the market within the window that the vendor wants
to provide us with.  My old Sun 2[1] at Interleaf was perfectly suitable
for running StunnedOS 3.5 and a (rather large) application.  The day
they drop support of 3.5 and the 2, I can't afford to use it for anything
but a personal machine!


| I think Rob Pike has the right idea:  put something positively doddering,
| like a 68010 :-), on the desks, and have it do *nothing* but interaction.
| Centralize the heavy computing where it can be updated centrally in a
| relatively economical way.

  I have to agree: the "2" would make a good X server, ditto a PC-XT, and
tritto an Xterm.  This means that I only have to stay within the window
for **hardware** maintenance, and can contract it out if necessary.
If the organization permits one to do so, this is a good choice.

  

--dave c-b
[1] the Sun 2 is in fact a 68010: this is the only architectural
    statement in the whole posting...
-- 
David Collier-Brown,  | davecb@yunexus, ...!yunexus!davecb or
72 Abitibi Ave.,      | {toronto area...}lethe!dave 
Willowdale, Ontario,  | Joyce C-B:
CANADA. 416-223-8968  |    He's so smart he's dumb.