[post 9/7/2020 9:02 AM 9/7/2020]
Post by Thomas KoenigClearly, in the 1980s, mainframes ruled the commercial world, at
least for large companies.
By that time, they had mostly lost the crown of scientific
calculation to the minis (with the exception of vector computers).
...
Post by Thomas KoenigSo, when did the mainframes start losing the commercial computing
centers? Is around 2000-ish the right timeframe?
Not exactly. A bit of history to put things in context
[in this writer's opinion and experiences]
Minis were certainly popular. But just like the Volkwagon
Beetle was popular, they had their limitations. Many
times researchers found their lab PDP was simply too
underpowered to handle their task and had to turn to
the corporate data center's IBM mainframe. Lo and
behold, the mainframe ran through a Fortran job in
no time. Often the mainframe programmers discovered
the researchers' Fortran program was poorly written
(e.g. unnecessary arithmetic statements within a
DO LOOP or unnecessary loops or I/O) and with a few
twists sped it up.
Post by Thomas KoenigSo, when did they lose their dominant commercial position?
Mainframe computers were and are heavy duty
machines. They can do a lot of work, but they
are very expensive. To be economically justified,
a mainframe needs a lot of volume.
IBM learned this when it found that its System/360
low-end models, even the budget model 20, was still
too expensive for small users. So it developed
the System/3.
Anyway, regarding the transition to smaller machines,
this happened gradually over a long period of time.
There was no magic date that mainframes "lost out".
In many organizations, the big IBM mainframe is alive
and well. IBM continues to upgrade its Z series
as we can see in the continuing new features described
in the latest Principles of Operations.
Certain aspects of the personal computer enabled it
to efficiently offload some of the work that the
mainframe used to do. For instance, accountants
would download raw files from the mainframe onto
their PC and use Lotus to do easily analysis.
This freed them from having a programmer write
and revise COBOL programs to research various
scenarios. As PCs grew in sophistication and
became more networked, their usage grew and
more work could be offloaded to them.
But I would posit there was also a 'herd mentality'
in corporate decision making. This has been
documented in the literature for the 1960s--a
lot of companies got computers because it was
the thing to do, not whether it would truly save
them money or improve efficiency. The CEO of
Avis wrote about this. There were many, many
computer system screwups in those years.
The CEOs demanded too much--the hardware
AND software of the era were stretched too thin.
(Geez, I remember when a bank put in a computer
system. They were so proud of it! But then customer
lines went out the door and around the block due to
delays. Finally they had to bring back
the old Burroughs posting machines less the
bank collapse, and rework the computer. The
same thing happened later to a department store
when it installed computerized registers. I
remember a truck bringing back the old
NCR registers and the salesclerks breathing a
sigh of relief. This sort of thing happened
a lot.)
Now, let's fast forward to more recent years.
Suddenly, the mainframe was referred to as
"legacy" or a "dinosaur", as were mainframe
programmers. "Client/Server" became the rage.
Why was that? Was it based on sound economic/
feasibility studies? Could Client/Server do
the work more effectively and efficiently?
In reality, there were four things going on:
First, IBM had lost its way. It had become
big and bloated and that meant its products
weren't keeping up with price/performance
demands of its customers. IBM had to get
an outside CEO and go through a major
contraction. Lots of people lost their jobs.
At the same time, some corporate data centers
became big and bloated* as well, and not as
responsive to user needs as they should've been.
Software development work became too bureaucratic,
took too long, and was too expensive.
The mainframe was seen as bloated, not an
image that was endearing.
Second, there were some tasks client/server
could do better than the mainframe. In
some cases the personal computer offered
cheaper hardware. In some cases it was
easier to program. Indeed, the mainframe
carried many 'baby' systems because
originally that was the only machine available.
When the PC came out, the baby systems could
be offloaded to it and were.
Third, the client/server world was young
and very aggressive. They were pushing their
product lines very hard. They exploited
IBM's troubles. The trade press touted
them. Having PCs do the work became all
the rage. CEOs and user department heads
didn't want to be left behind or be seen
as old fashioned. The Internet, despite
its many problems (like security) became
something to be worshipped. Having
access was a status symbol.
Fourth, there were corporate politics at
work. In many cases, the new client server system
was under the control of the user department,
not the central data processing department.
Department heads wanted the control--and
power--of owning and running the computer
themselves. That was extremely important
to them, rightly or wrongly. It had
nothing to do with efficiency, rather turf.
Indeed, often it was wasteful and a screwup**.
This power grab certainly help push the
transition away from the mainframe.
* Some of the bureaucracy was unnecessary,
imposed by clueless corporate auditors.
(Read an auditing textbook and you'll be shocked
at the b/s they recommend).
However, some of the bureaucracy was indeed
necessary to protect data files and software
from accidental deletion or deliberate sabotage.
Many of the client server sites had major
screwups that never would've happened in
traditionally run mainframe sites. They
had to learn the wheel all over again.
** Just as mainframers had to bail out the
research labs when their little PDP couldn't
do the job, mainframers later had to bail out
client/server sites that got screwed up.
So often a 'wonderful software product' proved
in practice to be not so wonderful.