From owner-freebsd-hackers  Thu May 15 13:55:56 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by hub.freebsd.org (8.8.5/8.8.5) id NAA14608
          for hackers-outgoing; Thu, 15 May 1997 13:55:56 -0700 (PDT)
Received: from fyeung5 (netific.vip.best.com [205.149.182.145])
          by hub.freebsd.org (8.8.5/8.8.5) with SMTP id NAA14601
          for <hackers@FreeBSD.ORG>; Thu, 15 May 1997 13:55:52 -0700 (PDT)
Received: from fyeung8.netific.com (fyeung8 [204.238.125.8]) by fyeung5 (8.6.11/8.6.9) with SMTP id CAA28059; Thu, 15 May 1997 02:04:35 -0700
Received: by fyeung8.netific.com (5.x/SMI-SVR4)
	id AA10443; Thu, 15 May 1997 14:02:06 -0700
Date: Thu, 15 May 1997 14:02:06 -0700
From: fyeung@fyeung8.netific.com (Francis Yeung)
Message-Id: <9705152102.AA10443@fyeung8.netific.com>
To: terry@lambert.org, jlemon@americantv.com
Subject: Re: Cluster Computing in BSD
Cc: rcarter@consys.com, pgiffuni@fps.biblos.unal.edu.co, fenyo@email.enst.fr,
        hackers@FreeBSD.ORG
X-Sun-Charset: US-ASCII
Sender: owner-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk


Greetings,

	These companies are so called supercomputer computers had no
intention to relate their business to any "day to day" computing. Their
machines were very target oriented and proprietary. 

	On the other hand, a cluster computing solution may solve a lot
of our scalability problems. What if our FreeBSD users want to enhance the performance of 
their Web servers, xxxDES computation, database applications etc.
An inexpensive cluster solution like the hypercube type will be very useful.

	You can get 100Mbps Ethernet NIC for $35 today. If the ATM
folks don't fool around with their hypes, we may see an inexpensive
OC12 ATM NIC soon (I may be dreaming too soon). 

If we believe that the local network can be the I/O bus, the clustering solution
 based on generic networking technology is worth for consideration.


	my 2 cents.

	Francis

> > I think the list of successes so vastly outnumbers the list of
> > failures that your parenthetical "nearly" is *way* out of place
> > here.
> 
> Hmm... let's see failures:
> 
>     Denelcor
>     Kendall Square Research
>     Multiflow
>     Cydrome
>     CDC
>     Convex		(aborbed by HP)
>     Burroughs Scientific
>     Scientific Computer Systems
>     Floating Point Systems
>     Supertek		(bought out by Cray Research)
>     Alliant
>     Myrias
>     Tera		(okay, so not dead, but not producing anything)
>     ...etc
>     
> This doesn't include companies that attempted to sell parallel computers,
> but eventually shut that division down, eg: Evans and Sutherland, BBN,
> and others.
>        
> Cray Research is effectively gone, having been absorbed by SGI, right?
> I'd say that Thinking Machines is on it's way out as well; the CM-5 is 
> obsolete, and I don't see a replacement on it's way.  Goodyear isn't 
> still building MPP's either, AFAIK.
> 
> That's not a long list of sucesses.
> --
> Jonathan
> 
>