LWN.net Logo

A new Linux/Windows TCO study

A new Linux/Windows TCO study

Posted Feb 14, 2006 3:35 UTC (Tue) by foo-bar (guest, #22971)
Parent article: A new Linux/Windows TCO study

Gents, do you really believe in all these funny studies ?
With all these funny precise numbers like 88%, 97% etc ?
Let's be realistic: IMHO all these studies, regardless of
which side they take are worth nothing. IMHO.

Let's concentrate on servers as this particular study does.

There two very different markets: big corporations with
professional sysadmins running thousands of machines of
diverse architecture and operating systems and, on the
opposite side small businesses.

The former are not worth discussing much, they typically
know how to count their dollars and usually quietly run
Linux servers for various tasks. Either way typically they
have well paid but at the same time highly professional sysadmins
there who can efficiently (with low overall TCO) run ANYTHING.
TCO is close to *ZERO* (per capita) anyway.
Even if they run Symbian on their servers :-)

With small business the story is the opposite.
In this world of high salaries for a group of 10-20 people
*ANY* PROFESSIONAL in-house IT service is cost prohibitive.

Now there are several options.
(1) Some sort of outsourcing. In this case the service company
runs everything and TCO estimates are irrelevant, you just
pay a (more or less) flat bill.
(2) Hiring a youngster - part time job. Not so bad but many
companies won't go for this for security/stability reasons.
(3) we are smart and we run our IT system ourselves :-)
Well, now let's estimate TCO.
(a) all employees are clever and smart professionals but with
"standard" Windows experience. TCO of running Lunux will
be equal to *INFINITY*. Believe me !
(b) There is somebody who knows how to type 'yum update' and
RTFM if (s)he needs to configure Postfix.
If (s)he is not stupid there is a good chance of TCO=0.

So, how can you estimate TCO if it ranges from 0 to Infinity
just depending on personal experiences ?

---

Now some personal experience. Which is actually a Linux TCO=0 case.

We, a group of 15 employees of not-so-small (but not-so-big :-)
company rent a virtual (XEN) server from unixshell.com.
We use it for virtually everything: mail server,
official website for customers, CVS server,
internal ssh/web server for file exchange, etc.

You know how much it costs per month ?
Half a normal dinner price :-) USD 30/mnd :-)))

You want to know TCO ? It's so small that nobody
bothers to reflect it in the timesheets. Roughly once a month
I do security updates when drinking coffee and I recall
I recall I spend ONE HOUR configuring Postfix a year ago
(which I never reflected in the timesheet because I couldn't
recall the date).

So in this particular case TCO ranges from 0 to USD 30/mnd
depending on whether remember or not to reflect the
monthly credit card charge on the expense report :-)

Now imagine that I was not in the group. What's the Linux TCO then ?
Well, it would be cost prohibitive then :-(


(Log in to post comments)

A new Linux/Windows TCO study

Posted Feb 14, 2006 20:47 UTC (Tue) by Just_call_me_fred (guest, #35163) [Link]

Agreed, this is exactly why Linux/Windows 'TCO studies' are useless. If you compare security, stability & configurability then that's more informative.

Copyright © 2013, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds