News from PANUG/BizNix - December 9, 2002
http://panug.org - http://biznix.org


NATIONWIDE WIRELESS PROJECT

AT&T, IBM, and Intel are backing a nationwide broadband wireless
project proposed by Cometa Networks. The goal is to offer a single
sign-on, single authentication, seamless-roaming nationwide network.
The service would allow anyone access without changing their accounts
or ISPs. Cometa plans to roll out the service in several cities by
late 2003. AT&T, IBM, and Intel will contribute to the project by
providing technology, either in the form of support or physical
resources. AT&T will contribute the networking infrastructure and
management; IBM will provide on-site installation support, and Intel
will deploy its new wireless-specific processor, code-named Banias,
due out in early 2003. 


SESSION C AT ITEC

Session C at the Linux Live event at the Itec show this Wednesday
will be presented by Red Hat Software and will focus on Red Hat's
plans for their future products. If you plan on attending any of
the Linux Live sessions, please register. Information on Linux
Live is at http://www.biznix.org/linuxlive/


WINDOWS/LINUX TCO
by Scott Hoffman

Here's a news item that bears including in the next newsletter:
http://www.crn.com/38852.html.  

According to the report, the total cost of ownership (TCO) for
Windows 2000 Server is 11-22% less than a Linux server over a five
year period.  Why?
Support staff and downtime are listed as the major contributors to
TCO (>85%), while software acquisition cost is less than 5%. Does
this mean that Linux requires more or higher paid staff?
Or that Windows 2000 Server has less downtime than Linux?

I'd be curious if one of the noted authors in the group (are you
listening, Ed?) would be able to get CRN to disclose their analysis
methodology and do a little peer review.

One possible explanation: I've commented before that Windows is very easy
to install, "out of the box". (I DIDN'T say it was SECURE, I said it was
EASY.)  Linux, in contrast, does require a certain amount of tuning and
configuration.  This might lead a small office without any support staff to
chose Windows 2000 Server.  Depending on how such deployments were treated
in the analysis, this could skew the results.

Ed Sawicki responds:

Yes, I'm listening. Note that my comments will be limited to servers.
This is IMPORTANT: I'm not commenting about Windows versus Linux on
the desktop, which is an entirely different issue.

TCO is an important issue. If, as you say, downtime is a major
contributor to TCO, then clearly Linux is the better of the two.
Unless a Linux SERVER is misconfigured or has hardware problems,
failures are rare. Server failures are expensive for many reasons
and often greatly impact TCO. Choosing an unreliable server
platform is an expensive mistake.

(Failures of a desktop computer are less expensive.)

Support costs are hard to measure. If you deploy Linux in your
server room but you know little about Linix, your initial costs
will be higher, of course. Training costs and consultants might
contribute to this. However, Linux support costs differ from
Windows support costs in at least one important way.

The knowledge you gain learning about Windows has a finite
shelf life. When the operating system changes significantly, as
it did from Windows 9X to Windows NT and again from NT to W2K,
there's a steep learning curve. Some or much of the old
knowledge is not portable to the new operating system.

Although Windows evolved from DOS, few of you still use your
DOS knowledge these days. You probably haven't edited a config.sys
or autoexec.bat file in years. Current versions of Windows evolved
from early versions of Windows but you probably haven't looked at a
win.ini or system.ini file in years. This has become throw-away
knowledge. Knowledge that you likely paid to acquire sometime
in the past. What's the return on investment?

When you acquire knowledge about Linux, Unix, or popular open
source software, the shelf life is far longer. If you learned
about Unix in the 1970s, most or all of that knowledge is still
useful today if you run Linux. 30 years later your investment
is still paying off. Support costs? I think Linux is a far better
investment and yields a far lower TCO - especially if you're
looking at a window of more than 5 years.

The report cites a savings over a 5 year period. Although less
significant to TCO, there's the issue of hardware costs. Within
5 years, you'll likely spend money replacing your Windows computer
with a more powerful one in order to meet the demands of a more
bloated Windows operating system. Linux operating system bloat is
far, far less (by an order of magnitude at least) and you can
recompile it to eliminate the code you don't need. New versions
of the Linux operating system don't force hardware purchases.

I'm growing weary of these silly reports, regardless of whether
Windows or Linux is the winner. It's rare that these reports are
objective in any meaningful way. I don't trust any of them. Even
those that report Linux as the better OS are sometimes suspect.

As IT professionals, it's best that we use our own skills to evaluate
products and technologies. Our trade press can't be trusted. The truth
is too easily purchased. Rather than reading articles like this, come
down to user group meetings and take advantage of the opinions and
insights of your peers. 

Remember, I've been commenting on Linux versus Windows TCO for
SERVERS - not desktops.