The debate on what per say is a mini...

From: Clint Wolff <vaxman_at_uswest.net>
Date: Sat Dec 16 13:13:40 2000

On Sat, 16 Dec 2000, Chuck McManis wrote:

> At 11:35 PM 12/15/00 -0700, Clint wrote:
> >I hate to jump into a philosophical argument, but here goes....
>
> Oh they are the best kind, everyone can be right at the same time ;-)
>

No, only me :)

> >Mainframes are big/powerful machines that sit in a secured area and
> >are driven by terminals (X,PC,3270,etc) on peoples desks. They cost
> >big bucks.
>
> In the "modern" world, what about an Intel server that is locked up in a
> co-location facility at some ISP and is being driven by web browsers ?
> (some of which (netpliance) are effectively terminals?)
>

I'd still consider it a micro. Most similiarly configured machines
are not supporting corporations, just individual users. The fact
that one is locked up doesn't really change it's status as a toy...

> >Micro-computers are small/weak machines that sit on peoples
> >desk so they can surf the web and play solitare. They cost
> >next to nothing.
>
> My PC sits under my desk, only the monitor/keyboard is on top :-)
>
I've got four sitting on my desk, each with a specific task to do :)

> >Workstations are more powerful, but still small... They cost
> >more than micros.
>
> Cost only? When I worked at Sun we worked in something like video
> resolution or max memory, but the bottom line has traditionally been cost :-)
>
Interesting point... Workstations typically have had better than
average video and memory capabilities than the typical PC, but
the line has blurred quite a bit... At work I have a Compaq P3-700
with 256MB ram (expandable to 1GB), and a Sparc Ultra 10-440(?)...
The Compaq is ~2x the speed of the Sun, but the Sun was designed to
use PC parts to reduce the cost, making it just another micro IMNSHO...

> >Mini-computers is a nitch that has gone away (or been replaced
> >by workstations). They were less powerful than a mainframe, but
> >orders of magnitude cheaper and could sit next to a persons
> >desk. This was more of a marketing trick to help scientists
> >convince the bean-counters they weren't buying a real computer,
> >just a glorified calculater (PDP == Peripheral Data Processor).
> >A calculator doesn't need to be locked in a room with all the
> >other calculators :)
>
> Actually this isn't a fair description. Mini-computers most definitely have
> _not_ gone away and you will see a resurgence of mini-computer like
> features. Specifically, mini-computers are computers that are
> *configurable* with a wide variety of peripherals. This lets them be used
> for lab experiments, or process control, etc. PC's have done this in the
> past but the lack of I/O features on the mainstream "PC" of today are
> hindering their use in the lab and process control. Also the lack of
> documentation that many people have complained about.
>
We use PCs almost exclusively for lab work... Of course, we build
custom hardware to do the work and the PC runs matlab to collect and
display the data...

> Collectively this is the "cost" model and generally follows order of
> magnitudes:
> < 4K Micro
> < 40K Workstation
> < 400K Mini
> < 4M Mainframe
> < 40M Super
> I added the last bit because we haven't been including "what's a
> super-computer" in our lists. I heard a great talk by a guy who works at
> LLNL and the complaints the scientists had at the purchase of ASCI WHITE
> (super computer made out of parallel RISC chips) because it simply could
> not be used to program the kinds of models they used for nuclear simulations.
>
> >PS It is unfair to compare mainframes of old to PCs of new. To be
> >fair, compare an 11/780 with what was available at the time: the
> >Apple ][, Commodore, IBM PC(?). These machines couldn't support
> >multiple users, imagine running a university off of a cluster of
> >Amigas!
>
> But Clint, this is in fact the point. Can one come up with a definition
> that transcends what the marketroids called micro, mini, and mainframe of
> their day?
>
Proly not, but it's fun to try...

> >PPS A modern mainframe would have no problem out-performing
> >a PC for REAL work. It's designed to do just that.
>
> Agreed, check out the spec's for the IBM S/390 (I mean Z900 :-)
>
> --Chuck
>
Received on Sat Dec 16 2000 - 13:13:40 GMT