!Re: Nuke Redmond!

From: allisonp <allisonp_at_world.std.com>
Date: Sun Apr 9 15:02:04 2000

>Yes, KO thought the DEC product line should be marketed to professionals.
>However, at the time, the people to whom a computer product had to be
>marketed was the hobbyists. It had been amply demonstrated that, no matter
>how marginally it fit, the personal computer as marketed to hobbyists would
>"do" in place of the supermini, with the trend toward distributed

PCs were useful in that they were local and flexible. reliability was hard
come by with them.

>spreading wildly, while no one could replace the 1000 or so PC's that cost
>what a supermini cost with a supermini. Professionals were, themselves,

True, by '87 the price/perfomance crossover for the individual user
was comming around.

>facility where my neighbor works. Their Pro-380 didn't do so well next to
>PC's costing less than half what they did. The high cost of DEC software
>licenses didn't help either.

By time the Pro380 hit, it was too late. Still it had things that PCs were
trying to do often not well. The license issue was costly but we are paying
for the alternative still.

>About 15 years ago, I was put in the position of demonstrating that a
>cluster of '386 PC's would outperform a custer of microVAXen in a given
>environment. What brought down the house was by how much they outdid them.

I did the same thing using z80s. What the vax did better was network, do
general applications and groupware. It was (even as a cluster) and still is
far easier to manage than 40 W95 PCs and 3 NTservers! I never achieved
the generalized performance of a vax even with multiple z80s.

>I was not nearly as sure of myself about that comparison as I had been in
>the previous SCSI/ESDI comparison. In fact, because of the substantially
>more efficient use of mass storage in the DEC MSCP, I expected that the
>would be I/O bound to their single hard drive, while the uVAX with a drive
>pair could operate much faster. There's quite a difference between what
>they can do and what they will do, I guess.

I found I could bring a PC to it's knees IO wise faster than a loaded vax
when the PC folded it wasn't graceful... still isn't under MS anything.

On the otherhand I've tuned vaxen and there are tricks that PCs still
learned. Try running multiple MSPC and mutiple SCSI controllers say two
of each with one spindle per... then Qbus is the limiting element. PCs
of the era of Qbus were ISA16 and they could never stand two EDSI and
two SCSI controlled in the same box, assuming you could get Win3.1
to even install them. Of course the 8-10 mhz bus was the bottleneck and
the controllers were not smart at all if they even had DMA.

Reminds me of the guy at citicorp in 1990 telling me his 386dx/33 could
crush the VAX750 (no racehorse). I wasn't impressed. He stopped making
the claim of superiority when I asked when the last reboot was due to a
Seems the vax was an unattended server with 1yr 3months uptime and was
also used to serve out his local PC database.

Comparisons are like many thing statisical... mostly point reference.


My $02 on PCs is Linux/FreeBSD/netbsd was the best thing to happen to
 PC hardware as it was the first OS with some semblence of performance
and concept of operational stability.

If there is anything worng with PC vs VAX comparison is often the vax
is nowhere near bleeding edge and the PC is. Then again the VAXs in
had more uptime than the PCs time in production. Also the VAX (running
OS of choice) had better IO buffering than most very highend PCs until the
early 90s.
Received on Sun Apr 09 2000 - 15:02:04 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:32:40 BST