[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Lucent Technologies & Sun Microsystems



>> There's at least one additional point.  The good solution today also should 
be scaleable to essentially arbitrary size, and should allow adding capacity in
small, affordable, easily-tuned pieces... and with a minimum of (or better, NO) 
disruption to the system components and procedures already in place.  Also, the 
effects of failures (hardware and software) should be small and bounded, much 
the same way that good programs ought to use a series of relatively independent 
and self-contained modules which deal with each other through a few simple, 
well-defined, inherently reliable arms-length interfaces.

>Absolutely!!!! I agree 100%. Us Bellheads know a lot about scaling and
reliability, etc. but our solutions are passe. 

Perhaps.  I certainly agree that telephone systems are in general superb 
examples of scaleable modular architecture and with (usually) bounded effects of 
failures.  I'm not sure it's so much that your SOLUTIONS are passe, or that you 
need to now apply similar understanding to radically different (and perhaps even 
MORE INTERESTING) problems.  

For one good example, take a look at the "Replacing the Automobile" project 
described on my Web page.  Here truly is **the** greatest commercial opportunity 
of our lifetimes.  It's a case where such scaleable, distributed routing, 
modular, real time network control architectures based on a humongous network of 
DSPs (and all tied together with Inferno, why not?) could *literally* change our 
world... and one could do it without butting head-to-head against Bill Gates.  
You want to REALLY do something cool with Inferno??  THERE is your project to 
use Inferno for... THERE is your showcase.

Yeah, I'm throwing down the gauntlet.  ;-)))

Let's take on a TRULY interesting (and hugely profitable!!) problem... one a 
great deal more compelling and important (and with a LOT more cash ultimately 
behind it!!!) than just scheduling my appointments from a distance.  :-)))  
Besides, it's a great deal more suitable as a project for an institution like 
Bell Labs and Lucent to do something that will TRULY change the world, shape our 
collective worldwide future.  

>We need to think like
netheads and make networks scalable and highly available using
techniques like sparing and dynamic routing. Adding boxes here and
there, not massive hardware upgrades.

Absolutely.

>If you go the modules route, then modules can exist on different
machines and cooperate -- that is Network Computing to me!

Good.  Now let's get rid of the so called "NC" foolishness that Oracle is trying 
to commit suicide with.  :)

Gordon Peterson
http://www.computek.net/public/gep2/