On Mon, 19 Jan 2009, James Tweet wrote:
> Ok, where should I begin. Let's start with the article's author. Did anyone=
> else look at his profile? He is "Originally a Math/Physics graduate who co=
> uldn't cut it in his own field." From reading this article I think it was =
> the math that gave him problems. To properly write an article like this you=
> should have a degree in finance, business or accounting.
Eh? That's like saying you shouldn't have an opinion on anything you don't
have a degree in. Hell, I'd never speak again. Hold on... I think I just
gave everyone here incentive to by in on that line of thinking. <G>
> =0A=0AArticle Quot=
> e:=0A"As far as I know the only way to make sense of this is to assume that=
> =0Ait took at least eight more people to run the 30 Linux servers than it=
> =0Adoes to run the 12 Windows ones -after all the money didn=E2=80=99t go f=
> or=0Areinvestment, it didn=E2=80=99t go for hardware, third party support, =
> or=0Asoftware, so what else is left?=0A"=0A=0AI can easily see how 8 people=
> would be needed to run 30 Linux servers. There would be 2 people per shift=
> just to deal with any problems on the Linux servers or network. Then you w=
> ould have a manager and an assistant manager. That would constitute the e-c=
> ommerce department.=0A=0AArticle Quote:=0A"As they found out, however, Linu=
> x isn=E2=80=99t Windows and changing to Linux on a one to one OS replaceme=
> nt basis only because it=E2=80=99s cheaper is about as dumb as it gets."=0A=
> =0AIf you can't replace Windows machines on a one for one basis. Then Linux=
> will never be a major competitor to Microsoft, period. =0A=0AIf using Linu=
> x would require a major infrastructure change in most I.T. departments. The=
> n most I.T. managers would not opt for a Linux solution.=0A=0AJames
Eh (Part II)? But he's right. Linux can certainly provide long term
savings, but your costs don't drop as soon as you overwrite a Windows
partition with a Linux partition. Linux expertise is more expensive than
Windows expertise, hands down. Unless you're real loose with your
definition of expertise, and since we've included Windows in this, perhaps
I'll have to concede that point. But the fact is that in general that more
expensive expertise gains you not only that knowledge, but a significant
increase in employee production, meaning that one Linux/UNIX admin can
typically manage many more servers than a Windows admin can.
On top of that, Linux provides savings by extending the life of your
hardware. But we're not talking by factors of a magnitude, in my practice
only by about 60%. So, that server lasts two more years, perhaps. You
still have to wait until *after* the normal lifespan expires before you see
any real savings.
I could go on, but this is the simple truth: regardless of what platform
you're going to or from, you will almost *never* see instant savings for
infrastructure *replacement*. In fact, the transition usually costs you a
pretty penny *above* normal operating costs, but you get to amortize that
over the lifespan of your platform and contract lifespans. With the lowered
*future* operational costs you can end up ahead. But not out of the gate.
And that's where the math of the case study is severely ethically-challenged
and misleading. It presents the numbers as if you can make a clean,
zero-cost break from one platform to another. And to that I say bull
puckey.
Now, you can make a case for one-for-one benefits when you're talking about
new infrastructure, but then that isn't remotely what the basis of this case
study was about.
That said, all the assertions that Paul makes are reasonable assertions.
The server "consolidation" is really a red herring, since the same
consolidation could have easily been achieved by upgrading the existing
platform's hardware running Linux, since they're comparing computing
capactiies that are magnitudes different. That alone wipes out the majority
of the cost savings. The remainder is easily eliminated as well once you
factor in OS licensing and uptime availability *if* you're comparing the
latest generation of MS software to the latest Linux.
The point Paul ulimately was trying to make, which perhaps could have been
made better, is that Linux itself isn't a panacea. If you're going to chart
a course into unknown an unfamiliar waters with no plan you're going to
shoot yourself in the foot. It doesn't matter what the numbers say on the
paper.
End sum: Linux can be done poorly, and for the wrong reasons, and it can
cost you. It's not the fault of Linux (normally), but the fault of
non-planning PHBs who don't know the right way to implement new/different
technology. And those morons gave MS the opportunity to spin the facts to
make an inferior product look good.
--Arthur Corliss
Live Free or Die
---------
To unsubscribe, send email to <aklug-request@aklug.org>
with 'unsubscribe' in the message body.
Received on Mon Jan 19 16:53:38 2009
This archive was generated by hypermail 2.1.8 : Mon Jan 19 2009 - 16:53:38 AKST