Subject: Scaleable Processing [was ...older, obsolete OS called "BSD Unix"...]
To: None <netbsd-advocacy@netbsd.org>
From: John R. S. Mascio <mascio@ryu.com>
List: netbsd-advocacy
Date: 04/23/2002 15:13:03
Mirian Crzig Lennox wrote:

> I'd never heard that (or are you kidding?)  I'd heard that NT was
> largely a VMS ripoff.


There have been accusations, but _I_ personally have not seen the 
proof.  But that is not to say that parts of VMS evolved in to 
parts of NT.


> But if it's true, then it's pretty neat that Win2K, MacOSX and *BSD
> are all, basically, cousins.  One big happy family.


Most of the current OS theory shows up in just about all modern 
operating systems, in one form or another.  Even the distributed 
OSs like Amoeba have the same components, but they have a 
different balance, implementation and design goals to achieve. 
We are still, mostly, constrained with the Von Neumann 
architecture.  Matter of fact, most people really can only think 
and design that way, either due to lack of mental ability or lack 
of education.  Not to say they are stupid, but limited in some 
fashion.

The major change w/ Mach is using message passing in a OS 
environment, which NT utilizes, but not as effective as Mach. 
Downside?  Your OS is slow due to the overhead of passing that 
message for a similarly functional, non message passing OS. 
Advantage?  The OS scales easily to more processors then the 
traditional monolithic design.

 From my limited studies, all the *BSDs have kept closer to the 
BSD 4.X model of single, multi-threaded monolithic kernels.  But 
then, only recently has decent, cheap multi processor systems 
been available to the general public.  Not to mention the modern 
personal computer environment has begun to allow individuals to 
have MANY CPUs in a home networked environment.  As the various 
BSD and Linux projects aimed at scalable multiprocessing evolve, 
I think we'll see more and more MACH message passing 
environments.  Some with a centralized control structure, some 
totally distributed.  I think most will fall somewhere in the middle.


I think the next generation of operating systems will take more 
concepts from OO, message passing and data flow theories.  But 
time will tell.

Coming back to NetBSD, being a newcomer, what projects are 
targeting scalable and distributed processing?

JRSM
-- 
       _  | John Raymond Stone Mascio
   _|_|_) | mascio@ryu.com
  (_|_|   | 214.725.7518
          | 972.240.5040
-----------------------------------------------------------------