Fog Creek Software
Discussion Board




shocking, misleading ad from Microsoft?

http://www.theinquirer.net/?article=18067

"The ASA said: "Because the comparison included the hardware, as well as the operating system and therefore did not show that running a Linux operating system was ten times more expensive than running a Windows operating system, the Authority concluded that the advertisement was misleading.""

(Required)
Wednesday, August 25, 2004

I believe it depends on the point of view.

If you compare them on the same hardware, you only take into account the cost of the software, and the cost of the people running it.  To do that, you've got to find a suitable control group with the same hardware but different software, and that's hard.

However, we know that hardware isn't free, or equally priced, or randomly distributed in conjunction with the software.  Software has it's hardware preferences, so to speak.

The model they probably used was to choose 'typical' hardware configurations for each software.  That way they could account for the cost of hardware in their TCO calculations, and the cost of hardware support.  In the end, a manager looking at the bill will want these figures too.

The question is, can somebody show that the typical configuration they used is not really typical?

Frank "Grimey" Grimes
Wednesday, August 25, 2004

Read the article - they compared Windows on a dual 900MHz Xeon machine to Linux on a two processor z900 IBM mainframe.

The standard Linux system is _not_ running on an IBM mainframe.

Using a dual-processor Xeon machine would have been a good comparison.

Using an IBM mainframe wouldn't since Windows can't run on a Mainframe...

RocketJeff
Wednesday, August 25, 2004

Did you read the article? It was comparing running Linux on a z900 IBM *mainframe* to Windows on a dual 900Mhz Xeon box.

A z900 is hardly a 'typical' hardware configuration for running Linux.

scruffie
Wednesday, August 25, 2004

IBM totes it as an ideal Linux enterprise server.  They're not going to compare it to a homegrown noname model.

Your posts suggest I'm crazy, but jeez, put forth some facts.  Just why isn't the mentioned model typical, and what model is?  Remember this is talking about enterprise markets.

Frank "Grimey" Grimes
Wednesday, August 25, 2004

By the way, I'm not condemning you or saying that Windows is better.  I'm just pointing out that the critic's argument is flawed, in that you can't expect the real world to have equal hardware configurations for different software.

<rant>
I really pity people who can't entertain a bit of objectivity.  Jump all over someone because they can't even accept the possibility that Windows *might* outdo Linux in something.  That's just sad.
</rant>

Frank "Grimey" Grimes
Wednesday, August 25, 2004

I didn't have a chance to read the original study.  But I don't think the cost of softwares has to be compared based on the same hardware, instead, they has to be compared under respective "best solution" configurations for similar application(s) with similar performance requirement.

XZ
Wednesday, August 25, 2004

It's like saying:
"an aircraft carrier costs more than a canoe".

Well, duh.

The point is, is *that* what you're trying to compare?  In order to have a viable comparison, you have to compare things with similar uses or within similar scenarios.

Therefore, a much more applicable comparison could be made between a canoe and a kiyak or between a battleship and an aircraft carrier.

KC
Wednesday, August 25, 2004

http://www-1.ibm.com/servers/eserver/zseries/900.html

vs.

http://www.swt.com/dualxeons.html

We're talking about IBM "Request a Quote" vs. "We can build it for you wholesale."

"Indeed, IBM has sold over 1,000 of its zSeries 900 --each worth well over $1 million-- since the product's rollout last year."
- http://www.cioupdate.com/news/article.php/10493_977831 (February 2002)

www.MarkTAW.com
Wednesday, August 25, 2004

Those links really drove the point home, marktaw- I was working on the same thing but you beat me to it.

funny thing- the Linux solution is about a 2,000 lbs (a whole ton!) heavier than the windows solution. 

This probably makes the first time a pure microsoft  implementation can claim an advantage in "portability" :P

PopCulture
Wednesday, August 25, 2004

The typical hradware for linux is a x86 processor. Which was Linus' original goal: a unix like OS on an x86 platform.

Is the ad "Shocking ?" No, the only thing "Shocking" is that people are surprised that an ad comparing 2 products is done in such a way as to favor one.

What would have been correct, would have been comparing linux on the same exact hardware as the windows, with the the default instalation. But then windows might have lost the cost war when adding CALS to match linux's client licensing.

anon-88
Wednesday, August 25, 2004

What I want to know is has windows EVER been put on a $1,000,000 box?

Seriously, can we get a count here?

www.MarkTAW.com
Wednesday, August 25, 2004

I ran Windows NT 4.0 on an Alpha 8500. It cost more than $1,000,000.00. I had several 4100s as well, and they were in the $500,000.00 range. They all ran like a champ and didn't blue screen periodically, or exhibit any of the other usual windows bashers complaints. Windows on Alpha was a beautiful thing.

Anomalous
Wednesday, August 25, 2004

Sorry, above should have read Alpha 8400...

Anomalous
Wednesday, August 25, 2004

>>What would have been correct, would have been comparing linux on the same exact hardware as the windows, with the the default instalation. <<

Again, I wouldn't agree with that.  If one OS performs poorer that the other, that wouldn't be a fair comparison either.

XZ
Wednesday, August 25, 2004

A Unisys  ES7000 can easily go over $750K. So I guess hitting one million is more than possible when adding the consulting fees.

anon-88
Wednesday, August 25, 2004

"What I want to know is has windows EVER been put on a $1,000,000 box?"


http://www.tpc.org/tpcc/results/tpcc_results.asp?print=false&orderby=os&sortby=desc

There are many. 

Jon Lindbo
Wednesday, August 25, 2004

>>Again, I wouldn't agree with that.  If one OS performs poorer that the other, that wouldn't be a fair comparison either.<<

Yes it is, then you would know how much more hardware (aka money) you would need to compare cost.

One reason I'm for stadard harware its and default installations is that you can tweek installs to get better perfomace. For example, it accepted that you can compile your own kernel in the linux world and optimize it for your specific processor.

On top of that, how do you compare performace? The number of static html pages served, gui performance, application performance,, the list goes on.

Besides, when comparing comparing things, isn't it better to eliminate variations?

anon-88
Wednesday, August 25, 2004

Reading between the lines, I'm guessing that Microsoft was really going after DB2.  Compared to Oracle and SQL Server, DB2 is a quiet giant - I think it has the #1 market share spot for the past 1 or 2 years.

When it comes to running DB2 on Linux, IBM is pushing the z800 and z900 series servers.  So, MS SQL on dual Xeons vs. DB2 on a Linux driven z900 might be a choice that some CIO manager needs to make.  Whether they would be swayed, that's another matter.

[For the record, I'm not advocating or defending MS on this issue.  I'm just speculating on the reasoning behind the comparison.]

Yet another anon
Wednesday, August 25, 2004

In the DB space, it might make sense, but the tasks being benchmarked were file and print serving.

Now who uses a mainframe to serve files? I buy a dual opteron now a day for this and with a few TB of RAID and a Tape Library, it's still cheaper then the Mainframe, minimal configuration...

Alexandre Carmel-Veilleux
Wednesday, August 25, 2004

>>On top of that, how do you compare performace? The number of static html pages served, gui performance, application performance,, the list goes on.<<

>>So, MS SQL on dual Xeons vs. DB2 on a Linux driven z900 might be a choice that some CIO manager needs to make. <<

I think that sums up all!  All should be decided in case by case bases. 

XZ
Wednesday, August 25, 2004

"There are many."

Thanks. Forgot about that list.

www.MarkTAW.com
Wednesday, August 25, 2004

IBM touts Linux as a scalable enterprise solution and, to a certain target market, does indeed use the example of mainframes.

It is a bog-standard comparison model for Microsoft enterprise marketing teams, and that's why they compared that model against Windows on Intel.

It's not misleading at all. Unfortunately, they presumed they could talk to enterprise marketing customers without worrying whether the rest of the world understood the ads, they way everyone else does, but forgot about open source fanatics.

.
Wednesday, August 25, 2004

The only thing shocking here is that MS pays hefty salaries to marketing managers who think that bare faced lying is a useful sales technique.

Mind you, considering that there are loads of brainwashed MS devotees like Grimes and . who would claim that Microsoft is quite right to say black is white because really everything is just a shade of grey, perhaps the marketeers have some excuse.

If you did a TCO comparison between American cars and European cars and compared a bog standard Ford with a Ferrari and then compared their performance for commuting and shopping in MYC nobody would pretend it was anything but junk. But GM. Chrysler and Ford were so busy making cars they forgot to set up a cult.

Stephen Jones
Thursday, August 26, 2004

Might I draw everyone's attention to this paragraph of the *actual* adjudication:

"The Authority noted the advertisers intended the advertisement to compare competing file serving set-ups that met the same needs and had the same function. It noted the study was audited and was a fair comparison of the operating systems on different hardware. The Authority considered, however, that because the advertisement stated " … WEIGHING THE COST OF LINUX VS. WINDOWS … Linux was found to be over 10 times more expensive than Windows … " it implied the comparison was between Linux and Windows operating systems only, and not about the performance of operating systems on different hardware."

The full adjudication is here: http://www.asa.org.uk/adjudications/show_adjudication.asp?adjudication_id=38475&from_index=show_advertisers&dates_of_adjudications_id=578

You'll note they're told to *amend* the advert. Summary: tut.tut, slapped wrist, go away & don't do it again. 

a cynic writes...
Thursday, August 26, 2004

> funny thing- the Linux solution is about a 2,000 lbs (a whole ton!) heavier than the windows solution.

I saw this story on /. yesterday (yeah, yeah, I know :-p ). They pointed out that you could run several instances of Linux on the mainframe against one on the PC. Short story: even the weight comparison isn't fair ;-)


Thursday, August 26, 2004

Microsoft expands mainframe pitch

http://news.com.com

.
Thursday, August 26, 2004

*  Recent Topics

*  Fog Creek Home