[comp.windows.x] Ergonmics of X-Terminals

mikem@telxon.UUCP (Michael Molnar) (10/30/90)

Our department was seeking justification for converting standard ASCII
terminal (GraphOn 230's) over to NCD X-Terminals.  Well, the gnashing of
teeth is over, upper management is convinced that we need them and the
check is as good as in the mail.  

But the justification did not come easy.  In this three part article I
document the saga of trials and tribulations.  In this part 1 of 3, I
acknowledge the replies from the users who candidly and generously responded
to my initial plea for help.  

========================== REPLY 1 =========================================
From uunet!flash.eta.cdc.com!tmorgan Fri Sep 28 11:10:42 1990
>
> Our department is seeking justification for converting its ASCII
> terminals over to X-Terminals.  Our past requests for X-terminals were
>

I'm not sure what "ASCII terminal" means.  However, if you are currently
running with monitors/software that does not allow multiple windows to
be used, here is an explanation that I have heard to help justify the
usefulness of windowing systems to brain damaged management.

Ask your manager to clear her desk of all paperwork, filing away everything
that needs to be retained.  Now, when any information is desired, only one
piece of paper can be taken out of the file drawer at a time.  Furthermore,
a template that allows only a portion of the page to be viewed must be placed
over the page and moved up and down and from right to left to allow viewing
of the information.  When additional information is needed, each piece of paper
must be returned to the files before the next piece of paper can be retrieved.
This is the equivalent of single screen terminal usage.

The first time I saw others running in a multiple screen environment, I couldn't
imagine how they could keep track of what they were doing.  With three years of
Apollo and SUN usage under my belt, I can no longer imagine working in any
other environment.

Good luck to you.

--
Toni Morgan               |"UNIX was never designed to keep people from doing
tmorgan@flash.eta.cdc.com | stupid things, because that policy would also keep
(612) 482-2634            | them from doing clever things."
                          |                  -- gwyn@smoke.BRL.MIL (Doug Gwyn)

============================== REPLY 2 ===================================
From uunet!sware!meo%sware.com.FORWARD Sun Sep 30 06:55:10 1990
|Our department is seeking justification for converting its ASCII
|terminals over to X-Terminals.  Our past requests for X-terminals were
|not approved.  The reasoning was: "Do you really need them?  Will they
|really increase productivity?  What's the pay-off?"
|
|Would anyone like to share their experiences with the change in 
|productivity which he or she achieved by switching an ASCII terminal based
|software group to an X-terminal base?  If possible, please detail
|with (1) the actual cost savings achieved by the conversion, (2) the
|type of applications used (please designate if they are X-only applications),
|(3) initial reasons for the conversion, and (4) the hardware environment
|for the X-window users.

My initial encounters with X were at my last employer. I plan to stay
with X a LONG time. Earlier encounters with Sunview (and somewhat with
Macs and Amigas) had already predidposed me towards windowing systems.
Since most of the rationale is the same, I will lump it together.

(1) The initial cost are higher, unless you use very expensive ASCII
    terminals. Typical Xstations cost from $2-3K, and a few are more.
    You may need extra disk space or memory on your system(s), depending
    on how many users you have and how much X software you plan to use.

(2) I develop software for a living - some of it X, some of it not.
    Interestingly, the payback is the same no matter what sort of software
    I'm working on. Most of the X software I use comes with X either
    from MIT or the system manufacturer. The rest comes from the net.
    Primarily I use xterms. I also use either mwm (Motif Window Manager)
    or twm (Tom's Window Manager - from the net or MIT). These alone
    justify the X terminal to me. Why?

       - I can see several files at once. I very rarely use printouts
         anymore for development, saving substantial paper costs, printer
         supplies (& wear & tear), and numerous trees.

       - I can monitor several things at once - two different compiles
         while I respond to email (elm is even better when you have 60
         or 70 lines to work with), write a memo, work on documentation,
         etc. I can easily bring up two different versions of a program
         and compare them side by side (without hunting down an extra
         terminal, cabling, etc).

       - I can cut & paste quite easily between xterms (and clipboards,
         etc.) with a mouse - much easier and faster than copying between
         files, jumping in and out of editors, etc.

       - It's quite flexible. I use large fonts for doing demos, and smaller
         fonts for getting more lines in a window. Most of the time, I'm
         working with 60-line windows, which are more readable than at least
         half of the 24-line ASCII terminals I've ever used.

       - I use xtroff to prescreen troff (typeset) documents without wasting
         paper.

       - Best of all, each xterm window can be attached to ANY SYSTEM on the
         net to which I have access. I typically have 7 or 8 xterms open at
         any 1 time, on 3 or 4 different systems. Our system administrator
         can do the same thing for all 30 or so inhouse systems, all from 1
         Xstation!

    I also typically use xrooms (developed at DEC WSL) instead of a session
    manager - it helps organize the "desktop" on my Xstation.

    I will be using xdbx soon - it's a very nice addition to dbx.

    I hope we can get an X-based spreadsheet soon. But meanwhile, sc
    (like elm) is greatly enhanced by large xterm windows (such as 70+
    rows by 20+ spreadsheet columns!)

    Finally, I monitor the system[s] with xnetload (if the system is
    getting very busy I tend to "nice" my compiles - if it's VERY heavily
    loaded, I know to do other work for a while!), get mail notification,
    and run xclock. I use the clipboard. Gnuplot is handy for occasional
    x-y plots, and the calculator is quite handy as well.

(3) Initial reason for the conversion was to develop software under the
    windowing system (originally Sunview, later X for portability, inter-
    operability with non-Sun systems, etc). The team using windowing systems
    had a proven productivity increase in a very short period of time.
    I have no hard figures for this, but would estimate 50 - 100 % for
    many tasks. The better the person is, the more they will be helped
    by moving to a windowing environment. This is the real payoff. The
    original $2-3K (plus running ethernet cable, twisted pair, or whatever)
    is quickly offset by the better use of the employees' time. For many
    shops (not just development shops), empoyee costs are the biggest
    portion of the budget.

    I now recommend the X environment for anyone who needs to do more
    than 1 task at once, or see more than 1 file at a time, or anything
    like that, who uses computers. Of course, if you need graphics anyway,
    you should consider X for the interoperability and portability. Even
    the stock market boys and the Fed are getting into X.

(4) My current X environment is the most complex of any I've used. We have
    30 or so systems, including Macs(A/UX), Honeywells (no X, but we window
    in), HP-UX 9000/3xx's, DECstation 3100s, Tektronix, Suns and Apollos
    (through a net to a local university), SCO-based 386 systems, and a
    few things I'm not allowed to talk about yet (seriously). At my last
    job we had Suns networked with a Convergent, dozens of 386 PCs, and
    a plethora of VAX/VMS systems, including workstations running X.

    We have several Visual X-19 Turbos and an NCR. We are getting an NCD
    and a Princeton/Computone for evaluation. Several people work directly
    on DEC, Tek or (the brands I can't talk about)  workstations, running
    X on the system monitors.

    I do NOT advocate X on PCs unless you have a large monitor with great
    resolution - 800x600 is just barely adequate if you need much on the
    screen.

Feel free to quote me in your net.summary, and in your justification. Also,
you might note that all the officers of this company, NONE of whom are
doing X development, feel the same way. I use an ASCII terminal at home
only because I have no choice, and am seriously looking at systems that
will run X (including SLIP-based Xstations) even for home.
----

Replying won't work if the From: or Return-Path: lines have .FORWARD in them.

Miles O'Neal
meo@sware.com  or  {uunet | emory}!sware!meo   (SecureWare)
{uunet | emory}!rsiatl!meo   (Systems & Software Solutions)

=========================== REPLY 3 =======================================
From uunet!watmath!watsol.waterloo.edu!tbray Sat Sep 29 15:45:03 1990
In article <544@telxon.UUCP> you write:
>Our department is seeking justification for converting its ASCII
>terminals over to X-Terminals.  Our past requests for X-terminals were
>not approved.  The reasoning was: "Do you really need them?  Will they
>really increase productivity?  What's the pay-off?"

If all you're doing is ASCII things, i.e., edit/compile/spreadsheet, etc.,
then your management is perfectly correct and X terminals are not worth
the *substantial* charge premium over ASCII.  If you want to do multiple
things at once, there are a variety of ways to arrange this on ASCII terminals,
e.g. emacs.

The only good reason to put a bitmap on a desk is if you have a working
application that uses bitmaps.  It is not currently the case that there
are so many such applications out there that buying the screen will almost
certainly pay off, as is the case with Dos or Mac screens.  However, there
are more and more such applications coming all the time, in the areas of
CAD, database, graphics, etc., etc...  The software company whose name
appears under my signature below is in the business of selling such things;
we all are waiting for the time when there's enough people like us that
bitmap workstation purchases are no longer controversial.

Cheers, Tim Bray (tbray@watsol.waterloo.edu)
Open Text Systems, Waterloo, Ont.

================================ REPLY 4 ==================================
From uunet!prcrs!wrwalke (and prcrs!nes)  Wed Oct  3 00:50:23 1990
In article <544@telxon.UUCP> you write:
|>Our department is seeking justification for converting its ASCII
|>terminals over to X-Terminals.  Our past requests for X-terminals were
|>not approved.  The reasoning was: "Do you really need them?  Will they
|>really increase productivity?  What's the pay-off?"

sorry about the length, but this is how i got my xterminals justified
to management.  when i came here, there were no Xterminals or workstations
at all (10 months ago).  in the last 3 months i have purchased 6 Xterminals
for my department (SAdmin), one for a loaner for demos, 4 for development,
and 2 for QA.  i have 6 more on order this month, and expect to have a dozen
or more ordered by Xmas.  i went from "no way, too expen$ive" to getting
a 21", 256 color, 4.5Meg RAM Xstation on my desk in two months just by
talking black and white $$$ (the management language) and showing an overall
savings.

here is the memo that got us started.  more to follow...

bill.


>XStation Justification

>Attached are the P.O.'s for six Xterminals.  These will be used
>by the five members of the Systems Administration staff (Shoemaker,
>Frise, Hite, Walker and Frost) who do a substantial amount of UNIX
>support.  The sixth terminal will allow us a "float" system
>to loan out to other groups for evaluation periods.  Long term,
>the sixth system will be used by one of the staff members in the Systems
>Support group of SA (e.g. Taylor, Fletcher) once site specific monitoring
>tools are developed by the UNIX support team members.

>A discussion of the requirements for the terminals and the alternatives
>that were considered is attached.  In summary, the productivity
>enhancements expected from these devices have contributed to our
>reduction in proposed staffing from 1 SA per 40 UNIX users in the 1990
>budget to 1 SA per 60 UNIX users in the initial 1991 budget.

>If you have any further questions, please contact me or Bill Walker
>who was responsible for the product evaluations.



++++>  here is my own personal justification of my increased
	productivity...

>i think if we put this into perspective, what would a MAC (yuck) or
>comparable PC product cost us with e-net bds??  how about sun or HP
>workstations??  our need is for multiprocessing (in the interface
>sense of the word) and multi-host access.  i can personally account
>for at least a 20% (conservative guess) productivity increase when
>working multiple hosts.

>Example:

>i am working on some code on prcrs.  in walks glenn and conor.  they
>need me to add an account on 4 hosts.  no problem, i open 4 windows
>from my hosts menu, add the accounts, and come right back to my code
>without ever closing my editor.  i don't like the response i am getting,
>so i glance down and see the load average running very high.  i open
>a concurrent prcrs window and run "top".  bryan calls with uucp probs
>at a remote site, i open another prcrs window and tail the uucp log.
>i open another window and log into the remote site to check it's
>load, # of users, and look around for abnormalities.  no need to log
>out and check uucp again.  by now i have been staring at a screen for 
>at least 3 straight hours and ${color:-"boy my head is aching"}
>${nocolor:-"i am sure glad i have color"}.



+++> and finally ... the dollars and sense approach...


>Let's assume that a GUI will improve productivity by 10%.  In particular
>all the users of these terminals require context shifts in response to
>user requests.  A terminal that can maintain multiple contexts and display
>information from more than one context at once will have immediate
>benefits.

>By providing the SA with responsibility for the non-networked systems
>an Xterminal we will not have to devote a full time person to the support
>of these systems.  Rather than spending the second half of that person's
>time on low leverage issues on the NCR systems, the person will be able
>to work in both the NCR and the HP/DEC contexts simultaneously and will
>be able to contribute to critical projects that span the hardware platforms.

>Finally, we expect to be able to produce graphics based monitoring 
>tools for the local systems that will allow the UNIX SA group to turn
>over some of this monitoring to the Systems SA group.  This will save
>another .5 UNIX SA.

>In summary then, the productivity gains are:

> 		3Q90	4Q90	1Q91	2Q91		
>NCR SA	 	.5	 .5	 .5	 .5	Frost
>Other SA 	.4	 .4	 .4	 .4	Hite, Walker, Frise, Shoemaker
>Tool dev       -.2				Hite, Walker
>Tool use		 .5	 .5	 .5	Hite, Walker, Frost

>		.8	1.4	1.4	1.4	

>This is a total of 1.25 man-years in the first year of use.  At
>the budget figure of $60K per person, this is $75K, approximately
>a 4:1 return on the investment in the first year.

>We've adjusted the 1991 budget to account for this.  The 1991 headcount
>in UNIX SA is 3 SA's for 180 users while the 1990 projected headcount
>was 4 SA's for 160 users.  The extra .4 person really represents
>monitoring work that we are balking now.   Proactive response to
>system problems will enable us to provide a more productive environment
>to all the system users.  This effect is not counted above.


>____________________________________________________________________
>Walker, Hite - HP/DEC - one 17" color, one 19" greyscale

>	Manage multiple contexts (problem report on one system
>	while working on performance issues on another) in multiple
>	windows.

>	Continuous performance monitoring  [display of graph of last hour of
>	system load on one or more systems]

>	Development of graphics based monitoring tools for
>	use by system support group.  Will eventually offload this
>	more routine work and allow the group to contribute in other
>	ways.


>Frost - UNIX/NCR
>	19" grey scale or 17" color

>	Critical need to manage multiple contexts since will spend 1/4
>	time supporting the NCR Towers without TCP/IP capability.  The
>	serial line of the XStation can be used to access one of the
>	Towers and the desktop can still be connected to the ethernet
>	capable systems.

>		Estimated value: .5 person immediately
	
>Frise - Systems
>	19" grey scale or 17" color

>	Need to monitor all systems.

>	Complex context including direct support for MLS products.

>	Requires access to a call tracking system to manage workload of
>	the group.

>	Produce reports including diagrams and tables - need a previewing
>	capability.

>Shoemaker -
>	14" monochrome

>	Preview reports without having to check hardcopy

>	Use drawing tool for diagrams for presentations

>	Continous display of analog clock -- time to go home...
	

>Sixth terminal
>	14" monochrome

>	Short term - spare
>		Lend for demos
>		Trade for larger or color monitor while that is
>			loaned out for demo
>		Use for cut and paste into "old" bugdb until Xchange
>			is in place

>	Longer term (4Q90) - system support console

>		Continuous performance monitoring on all systems
>		Call tracking system monitor
		

============================== END OF REPLIES ======================

In part2, I will present the qualitative arguments that our group
presented for X-terminals.  So, stay tuned to this newsgroup for
more "Ergonomics of X terminals".

-- 
Mike Molnar		|  uunet!telxon!mikem	   |    (216) 666-4300
TELXON Corporation	|-----------------------------------------------------
91 Springside Drive	|  "No, mom and dad, don't touch that!  It's evil!"
Akron, OH 44313		|             -- Time Bandits