Letter
To The Editor: Sun Linux Desktops
It
seems appropriate to print this Letter to the Editor in this
issue, since subscriber John McCreesh from the UK wrote it in
response to my original Flash on the Sun Linux Desktop
announcement on September 18th.
He wants to make two points:
-
Use
any old PC as a terminal if you want server-based Linux, not
someone’s specialty product; and
-
No need to buy
anything, all the parts are available for free.
Read
on:
Amy:
Sun
gets an A+ in identifying a user need for a sustainable computing
platform, but a C- in coming up with a plausible solution. Buying
dedicated thin terminals - even ones as sexy as the Sun Ray - is a
risky business. If the system doesn't work, you're left with a
pile of kit that can't be reused and which has zero resale value.
And less than successful deployments of terminal/server systems
based on Microsoft technologies has given terminal/server a bad
name.
This
is a shame, as there are much 'safer' ways for organisations to
experiment with / migrate to sustainable computing. Pretty well
any networked PC can be converted instantly into a dumb terminal
simply by booting it from a suitable floppy. Load Linux onto a
spare server, add the package of 'glue' from the Linux Terminal
Server Project (LTSP), and you can be up and running on a
sustainable computing Proof of Concept without spending a cent.
All the software is open source, and so free of license fees.
The
case for sustainable computing based on terminal/server and
open-source is pretty compelling (see below for the long version).
Sun's route is expensive and risky - the Linux/LTSP approach is
cheap and risk-free. Sun would be better packaging LTSP on
Solaris, thereby replacing Microsoft PCs with Linux diskless
terminals, and concentrating its expertise on offering highly
scalable server solutions. Once users got comfortable with
terminal/server, then Sun could start offering Sun Rays for new
desktops or as replacements for worn out PCs.
John
McCrees john.mccreesh@bcs.org.uk
Editor’s
Note: What follows is
what John McCreesh calls the “long version” – his remarks on
why users might want an alternative to current office software on
the PC platform. It’s
an argument for Open Source and the use of a particular piece of
Open Source software that’s designed to permit multiple
terminals to share a Linux server, offered by the Linux Terminal
Server Project. You
will find documentation on this project at http://www.ltsp.org/documentation/ltsp-3.0.0/ltsp-3.0.html.
Some of you may agree; others will be (I’d guess) in
strong disagreement. It
seemed like a good letter to print on OpenOffice.org’s second
birthday.
John
Mccreesh On The Need For Sustainable Computing
There
is a huge marketplace for users on a network who want basic office
software: email, browser, word processor, spreadsheets,
presentations. These users are mostly just ordinary folks who
maybe use 5-10% of all the functionality of their current
software, and don't need/want additional functionality.
Unfortunately, desktop PCs are an extremely inefficient means of
delivering computer power to where it's needed. Even a 'power' PC
user on a busy day uses no more than a few percent of the
computing power sitting on their desktop. And of course, as soon
as one user gets a shiny new PC, then everyone else 'needs' one...
It's
now widely acknowledged that keeping desktop PCs running is
expensive in support costs. Running multiple versions of operating
systems / applications software increases costs. Upgrading the
software on PCs is massively expensive, as today's software rarely
runs on yesterday's PCs (yesterday could be as little as one year
old). Getting rid of old PCs is expensive - wiping disks, removing
asset tags, complying with environmental regulations. Consumer
protection legislation and licensing restrictions means corporates
increasingly can't even give away PCs to charities
However,
Microsoft is committed to driving out new versions of Windows and
Office every three years to keep its shareholders happy, and
simply refuses to sell/support old versions. It's no accident that
three of the world's richest five people owe their wealth to
software (Forbes magazine Billionaires list 2002). This is a huge
incentive to keep the treadmill of enforced obsolescence turning
as fast as possible.
What
users actually need is sustainable computing - giving users the
computing tools they need - no more, no less - without costing the
earth. This means sustainable software and sustainable hardware.
The
only guaranteed route off the treadmill of vendor-enforced
software obsolescence is to change to open-source software under
an OSI licence. Open-source software for the desktop has improved
dramatically in the past few years (particularly with the release
of OpenOffice.org). Most Microsoft Office users can now move to a
Linux desktop with their existing documents etc. and hardly notice
the difference.
(Editor’s
Note: We’d argue
that this depends on what kind of documents they have.
Ordinary text documents will do pretty well, with some
kinds of advanced formatting not being supported; spreadsheets and
other applications definitely have their limitations.)
The
most effective use of computing power is to put an absolute
minimal hardware on a diskless terminal on the desktop, and do all
the 'real' computing on shared central servers. Virtually all the
PCs currently being thrown out by corporates could be used
indefinitely as diskless terminals. Comparatively small servers
can support surprisingly large numbers of users- simply because
the available horsepower is being used efficiently. Holding all
the user data and software centrally provides for an extremely
cost-effective support model.
So
why has terminal/server computing got a bad name?
Attempts
to produce client/server versions of Microsoft operating systems
have proved unsatisfactory for two reasons. Microsoft Windows -
and all applications based on it - are architected on assumption
that they will be run on a self-contained PC. Trying to split off
a terminal is an unnatural concept. Windows has also not been
conspicuously successful on the server as a multi-user
multi-application operating system, although it is gradually
improving.
Terminal/server
computing has also been largely driven by vendors of thin-client
terminals. Unfortunately, vendors have been unable to generate
sufficient volumes to drive costs below ordinary PCs churned out
in huge numbers. Buying dedicated thin-client terminals is a also
risky option - if the terminal/server system doesn't work, you're
stuck with a lot of useless equipment with zero resale value.
However,
away from the Microsoft camp, terminal/server is a much more
attractive proposition. Linux/*nix was designed from the outset as
a multi-user, multi-function server operating system. When
windowing systems were added to Unix, they were architected from
the start to be separate from servers - the complete opposite of
Microsoft Windows. You don't need to add anything special to Linux
to do terminal/server - it's already there. What has been missing
has been the expertise to do Linux terminal/server computing 'out
of the box' - with the work by the Linux Terminal Server Project,
a package solution is now available.
As
all the software required to try out Linux terminal/server is
available under open-source licenses, it's very easy to try out.
Pretty well any networked PC can be converted instantly into a
dumb terminal simply by booting it from a suitable floppy. Load
Linux onto a spare server - or even a good desktop PC, add the
package of 'glue' from the LTSP, and you can be up and running on
a sustainable computing proof of concept without spending a cent.
It's an interesting exercise to try and see what the users think
of the sustainable computing alternative. Reverting to Microsoft
Windows is as easy as removing the boot floppies - but the chances
are, you won't want to do it.
John
McCreesh
john.mccreesh@bcs.org.uk
(back to top)
Comments or Questions: Send Email to
opinions@wohl.com
Home/
Search
/ 2005 Articles /
Issue Archive / Free
Newsletter
Entire contents © 2001 by Amy D. Wohl. All rights reserved. Reproduction of this publication in any form without prior written permission is forbidden.
