Re: RE : Re: [tlug] Small footprint Linux distribution without a GUI

[Curt Sampson <cjs@example.com>]

On Mon, 14 May 2007, Stephen J. Turnbull wrote:

> The Coda list is full of people in pain because their application
> cannot stand up under NFS semantics, but their users refuse to "gaman"
> through Coda semantics. (They typically propose to reinvent NFS
> semantics, but "somehow" make them atomic like Coda.)

And they'll stay in pain, because they're searching for the
unobtainable. Distributed computing is qualitatively different from
single-node computing.

In single-node computing, such as with a stand-alone PC, any component
failure (such as a drive cable being unplugged) causes complete system
failure (everything running on the PC dies). This is acceptable to most
people because such failures are relatively rare, and it's only one PC
anyway.

In a more distributed system, such as with an NFS file server serving
all of the programs and shared libraries for many PCs, the failures are
much more frequent, and even expected (most sysadmins would be quite
surprised if unplugging a network cable for a few seconds and plugging
it back in caused a network collapse), and the system itself is rather
larger (people who would shrug at an individual PC going down become
quite irate when the whole office shuts down).

The only way to deal with this is to restructure applications to handle
the new circumstances gracefully, which often means throwing the entire
thing out and rewriting it from scratch.

BTW, Gregory Pfister's _In Search of Clusters_ is essential reading
for anybody seriously interested in or having to work with distributed
systems of any kind. If I had to restart computing civilisation from
scratch, this book would definitely be one I'd pack.

cjs
--
Curt Sampson       <cjs@example.com>        +81 90 7737 2974

Mobile sites and software consulting: http://www.starling-software.com