Subject: Re: [coallesced replies] Re: how to create 21,780 new free software jobs (2,530 in R&D)
From: Tom Lord <lord@emf.net>
Date: Sat, 9 Nov 2002 10:45:39 -0800 (PST)



	chrismaeda:

	I also wouldn't waste my time on the distributed file system
	R&D.  It's not relevant to the business model.

As eichin points out: I'm not up to speed on the state of the art in
this area -- but DFS seems quite relevant to me, because it provides a
very cost effective, technically flexible way to implement remote
administration, scalability, "availability everywhere", client device
independence, and ease of use.  This was proven at CMU in the 80s.  I
know of no other combination of technologies that hits those features
in such a simple, elegant way.

Many other approaches to remote admin have been developed for server
farms.  Many have been developed for Windows desktops.  My accumulated
sense is that these don't work as well, especially for the more
complicated kinds of environments we see outside of servers.  I would
be very leary of a duck-tape, rube-goldberg approach to doing this.

"My accumulated sense" isn't critical to the business model, but
historical facts are, so some survey research of the literature and is
needed.  I can't, with intellectual honesty say "certainly distributed
file systems are the way to go" -- but evaluating them seriously and
in detail certainly _is_ relevant to the business model.


	 You might consider doubling or tripling the price and
	 throwing in a computer.  It seems that a business like this
	 might have the same economics as a cell phone carrier.
	 Including the computer would also lower your support costs
	 because you control the hardware platform.

Absolutely: control over the harware platform is critical.  Computers
need to be more like phones: cheap and swappable and relentlessly
standardized in this area.  Is it a rental model we want?  A shelf of
endorsed hw at CompUSA?  a page at Dell's web site?

The cost/price analysis focused entirely on software.  It spun off a
$320k/year "facilities budget" which looks to me like one $190k
business administrator and a paltry $130k for capital outlays.
Clearly, the plan (as it stands) is for "business units" rather than
"complete businesses".

So, I dunno -- where's our mgt team? :-)



	robert:

	Many businesses and individuals are very sensitive
	to remote companies storing the kind of very
	confidential/personal data on a remote server,

There are two spaces where solutions there might come: (1) secure
remote file systems (and secure storage); (2) appropriate business
boundaries.  Regarding (2), if a large organization wants to possess
their own storage servers -- more power to them.

"Andrew Sucks" -- the Great AFS/butler-pool worm of 1987
-t