Subject: Re: on the horizon -- broadband filesystems v. distribution
From: "Forrest J. Cavalier III" <mibsoft@mibsoftware.com>
Date: Thu, 31 Oct 2002 11:17:36 -0500 (EST)

"Benjamin J. Tilly " <ben_tilly@operamail.com> wrote, in part

> http://www.stuartcheshire.org/rants/Latency.html

  My summary: The speed of light isn't just a good idea, it's the law.

> http://www.darkridge.com/~jpr5/doc/gnutella.html

  My summary: Ponzi was wrong/Chain letters don't work/Don't
              try to put 2^64 grains of rice on the last square
              of a chessboard.

Thanks for the links, Ben.  Can I add two other related
"basic engineering truisms"?
  - Your influence is localized.
   You have more influence on your local power company to fix an
   outage than a backbone provider.

  - Reliable complex systems are hard to implement.
   10 points of failure which are each 99.99% reliable means
   8 hours of downtime a year.

For that last point, consider that most commodity PCs do not
have 8 hours of downtime a year.  Most people's televisions
don't have 8 hours of downtime a decade.  In contrast, I
think I am prevented from doing what I want over the Internet
more than that (due to server and network outages, etc) and I
surely am not on the Internet 100% of the year.

(And extrapolating...if .NET is what I think it is, it isn't
going to be accepted widely anytime soon.)