Re: LINUX - Total Cost of ownership

On Fri, 14 May 1999 12:13:50 -0500, Brian Fristensky <frist_at_cc.umanitoba.ca> wrote:
>"Christopher B. Browne" wrote:
>> TCO calculations are *highly* politicized. They are probably about the
>> *least* scientific of calculations that anyone pays attention to.
>>
>> They are used as finely-tuned weapons in the war of marketing.
>>
>>I expect that trying to do a scientifically-oriented analysis will
>>have the primary effect of making your head hurt a lot...
>
>I agree. The only thing I can add is that the first time
>I heard the phrase "Total Cost of Ownership", it was used
>to promote the concept of Network Computing, or Thin Client
>computing. The idea is that, while PCs seem cheaper than servers
>it is cheaper to maintain a few servers and terminals/NC's than it
>is to maintain an equivalent number of PCs ("Fat Clients"). Not
>surprisingly, everyone, including (giggle) Microsoft,
>has started claiming that they have the lowest TCO.

The term has been used for quite some years now; it has only been recently that Linux was sufficiently "on the map" that there has been connection.

The big problem with TCO is that it represents an arbitrary determination and allocation of costs, and, even moreso than with benchmarks or "system reviews" (e.g. - Mindcraft, TPC-D...), the interpretation of the results may be interpreted and re-interpreted so as to come up with whatever result you *want* to come up with.

In the business "world," they evaluate TCO to try to come up with the most economical computing platforms.

It's obvious that you have to consider such costs as hardware and software licenses.

But *realistically,* you need to add in other stuff. - Support contracts
- Electricity (e.g. - the Corel Netwinder claims "lower TCO" due to using less power than a PC)

- The army of system administrators
- Help desk staff
- Help desk software
- Asset management costs
- Training costs
- Sometimes licensing costs get really complex
- Costs of retiring the system at the end of its life
- Costs of disposing of computer hardware
- Revenue from selling old hardware...

What will typically happen is that Novell will claim that Netware 5.0 has superiority to Windows NT 3.51 due to having a "30% lower TCO."

Microsoft will then complain that Novell ignored some area of cost, thus pulling in some more costs for both NT and Netware.

Sun Microsystems may then rewrite the analysis and claim a lower TCO based on other fiddling with metrics.

Throw in RDBMS vendors like Oracle/Sybase/Informix, or ERP vendors like SAP AG, Oracle, PeopleSoft, and add to the fun...

TCO thus tends to represent making up numbers that are only really usable in the context of a particular enterprise, and vendors have ample incentive to encourage throwing in those criteria that make them look better than their competitors.

Trying to extract any objectivity from this is hopeless.

-- 
Would-be National Mottos:
USA: "We don't care where you come from.  We can't find our *own*
country on a map..."
cbbrowne_at_hex.net- <http://www.ntlug.org/~cbbrowne/lsf.html>