Linux-Advocacy Digest #491, Volume #32 Mon, 26 Feb 01 05:13:07 EST
Contents:
Re: NT vs *nix performance ("Boris Dynin")
Re: M$ doing it again! ("Erik Funkenbusch")
Re: M$ doing it again! ("Erik Funkenbusch")
Re: RTFM at M$ ("Erik Funkenbusch")
Re: NT vs *nix performance (Klaus-Georg Adams)
Re: Microsoft says Linux threatens innovation ("Erik Funkenbusch")
Re: Microsoft says Linux threatens innovation (Ed Allen)
Re: Microsoft says Linux threatens innovation ("Erik Funkenbusch")
Re: Mircosoft Tax (Donovan Rebbechi)
Re: Mircosoft Tax ("Mart van de Wege")
Re: M$ doing it again! ("Edward Rosten")
Re: Mircosoft Tax (SoneoneElse)
Re: M$ doing it again! ("Edward Rosten")
----------------------------------------------------------------------------
From: "Boris Dynin" <[EMAIL PROTECTED]>
Crossposted-To:
alt.destroy.microsoft,alt.linux.sux,comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: NT vs *nix performance
Date: Mon, 26 Feb 2001 00:11:13 -0800
> > Check this one:
> > Tester System Name Result HTTP Version CPU #
> >
> > Dell PowerEdge 8450/700 7300 IIS 5.0 and SWC 3.0 8
> >
> > http://www.spec.org/osg/web99/results/res2000q4/web99-20001211-00082.asc
>
> Exactly - microsoft's "bet the farm", "benchmark buster"
> configuration with the special web cache in front of iis
> is badly outperformed by AIX, and can't even match the
> performance of the free Red Hat Linux system.
We discussed it on this group already. Linux results are for some strange
kernel mode web server called Tux which nobody seems to use. This means (to
me) that Linux results are quite useless.
As for AIX results, they were achieved on more powerful end expensive
hardware than IIS test above: 12 CPUs for AIX vs. 8 for W2k, 64GB RAM for
AIX vs. 32 GB for W2k.
>
> Quite a different story from the "windows is 4 times
> faster than the fastest Unix" crapola that was posted
> here earlier.
I wouldn't be surprised if AIX test configuration did cost several times
more than W2k.
>
> > > And oddly enough, a $6000 Linux server easily outperforms
> > > this suspiciously slow and overpriced "unix system", and also
> > > compares favorably with the $208,000 windows pc server farm.
> > And yor sources are...
>
> I have a very low end server (Compaq Pentium Pro 200)
> here at the office - I did some quick benchmark tests with
> apache, mind you - not something really fast like zeus or
> tux, just plain old apache on Red Hat 7.0,and I get 4200
> requests per second from that tired old Linux box, over
> a single100 mbit ethernet interface.
The question is what you were measuring. What makes you think that your
"quick benchmark tests" make any sense at all?
The rest snipped.
Boris
------------------------------
From: "Erik Funkenbusch" <[EMAIL PROTECTED]>
Subject: Re: M$ doing it again!
Date: Mon, 26 Feb 2001 02:52:56 -0600
"mlw" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> Erik Funkenbusch wrote:
> > Those API's are not intended to be used outside of the kernel. I can
> > reverse engineer the windows kernel too, doesn't make it any more
> > "published".
>
> I have a real problem with what you write sometimes. While we often have
> differences of opinions, you usually seem fairly reasonable, then there
are
> times when you will create a paragraph as above. You can't possibly
believe it,
> do you? You must say these outlandish things just to be an idiot.
>
> How can you say that open, "published," source code can in anyway be
> undocumented. In the linux kernel there is nothing that is undocumented,
one
> can see every single API and what it does. Just because someone didn't
dump the
> comments to a separate text file, does not mean it is undocumented.
A "published" API is one which intended for external use. There are plenty
of functions in the linux kernel which are not intended to be used outside
of the kernel itself. Those are unpublished, and undocumented since they
are not listed in man pages (unlike the functions which are intended to be
used externally, which do exist in the man pages).
------------------------------
From: "Erik Funkenbusch" <[EMAIL PROTECTED]>
Subject: Re: M$ doing it again!
Date: Mon, 26 Feb 2001 02:49:08 -0600
"Aaron Kulkis" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> Erik Funkenbusch wrote:
> > All OS's have unpublished API's, including Linux. You can find all
kinds of
>
> Wrong on two counts.
Really?
> > API's that are inside the kernel but do not appear in man pages because
> > they're intended to be used internally inside the kernel itself without
a
> > published interface.
>
> 1) If it's only meant to be use INTERNAL to the kernal, and
> *NEVER* by application, then it's *NOT* part of the API.
>
> API = APPPLICATIONS Programming Interface.
>
> Anything which is "internal to the kernal" is, by definition, outside
> of the scope of the API.
That's an awfully narrow view? What exactly do you call the kernel module
API? A kernel module is not an application, yet the kernel exposes a set of
interfaces for the modules to link to. Those interfaces are commonly known
as API's, even if it's not an application.
> 2) What part of OPEN SOURCE do you not understand?
>
> The source-code IS *ALWAYS* the authoritative documentation.
Not everything in the kernel is documented in source comments either.
------------------------------
From: "Erik Funkenbusch" <[EMAIL PROTECTED]>
Crossposted-To: alt.destroy.microsoft
Subject: Re: RTFM at M$
Date: Mon, 26 Feb 2001 02:57:14 -0600
"Jim Richardson" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> >I understand them perfectly. I guess, according to you, if internet the
> >backbone goes down, then the destination server is also down just because
> >you can't get to it.
>
> it's certainly not available, which is the practical equivilent.
Not available to you doesn't mean it's not available to others.
------------------------------
From: Klaus-Georg Adams <[EMAIL PROTECTED]>
Crossposted-To:
alt.destroy.microsoft,alt.linux.sux,comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: NT vs *nix performance
Date: 26 Feb 2001 09:41:52 +0100
"Ayende Rahien" <[EMAIL PROTECTED]> writes:
[...]
> I followed the trail of URLs and found the following comments:
>
> http://www.zdnet.com/eweek/stories/general/0,11011,409114,00.html
> "The results show that Microsoft so dominated the test--in terms of price
> and performance--that it looks downright embarrassing for the competitors"
>
> http://www.zdnet.com/eweek/stories/general/0,11011,409389,00.html
> "...its [microsoft] C++-based bookstore was so fast that it was bottlenecked
> by the 100M-bps test network."
>
> More information about this can be found here:
> http://www.zdnet.com/eweek/stories/general/0,11011,409380,00.html
>
> I know that this is quite old(almost 2 years), but is there some new data on
> it?
You might want to look at this for news about NT performance:
http://www.midrangecomputing.com/showcase/article.cfm?titleid=b1534&md=200011
<quote>
Dr. Frank Soltis, the IBM engineer who has been called "the AS/400's
Elvis," recently shared a success story during a keynote speech at a
user conference in Florida. This particular company was in the
software distribution business and at one point had 23 AS/400s located
around the world. The company was a very good customer, went from CISC
to RISC, and was always one of the first to upgrade to new technology,
he said.
Then came the Year 2000 problem, and despite five years of dedicated
service during a period of great revenue growth, the company decided
that it was time to move off the AS/400. So in June of 1999, the
company unplugged its AS/400s and powered up the 1,200 NT servers it
needed to replace them. But things didn't quite go as planned. "They
found they couldn't make it work," Soltis told the crowd. "Today, one
year after unplugging their AS/400s, they're back on the AS/400."
That company is Microsoft. "They viewed that as a point of
embarrassment," Soltis said. "We thought it was kind of fun....Can you
think of a company with greater incentive to move to NT, and they
couldn't do it?"
</quote>
Chuckle...
--
kga
------------------------------
From: "Erik Funkenbusch" <[EMAIL PROTECTED]>
Subject: Re: Microsoft says Linux threatens innovation
Date: Mon, 26 Feb 2001 03:09:47 -0600
"T. Max Devlin" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> Said Erik Funkenbusch in comp.os.linux.advocacy on Sun, 25 Feb 2001
> >"T. Max Devlin" <[EMAIL PROTECTED]> wrote in message
> >news:[EMAIL PROTECTED]...
> >> Said Erik Funkenbusch in comp.os.linux.advocacy on Sun, 25 Feb 2001
> >> >"Chris Ahlstrom" <[EMAIL PROTECTED]> wrote in message
> >> >news:[EMAIL PROTECTED]...
> >> >> Erik Funkenbusch wrote:
> >> >> >
> >> >> > Software is not hardware, and it doesn't follow the same market
> >trends
> >> >of
> >> >> > hardware. For instance, Adobe Pagemaker has stayed the same price
> >for
> >> >the
> >> >> > last 6 years as well. Does that mean Adobe is also monopolizing?
> >> >>
> >> >> The price of Microsoft software has not stayed level. It has
increased
> >> >> by an incredible amount. For example, about 3-4 years ago, I bought
> >> >> an OEM version of Office for about $150. The same version is now
about
> >> >> $500. How's that for monopoly practice!
> >> >
> >> >Not even.
> >> >
> >> >http://www.pricewatch.com/1/146/1904-2.htm
> >> >
> >> >Office 2000 Standard OEM costs roughly $175, and that's through a
> >middleman.
> >> >If you buy directly from the distributor, I believe it's about the
same
> >> >price as you claim.
> >>
> >> So he's right. Was that your point?
> >>
> >> (For future reference; all 'street price' quibbling will be
> >> appropriately ignored entirely. List price is the only price that
> >> matters for this discussion.)
> >
> >You can only buy OEM copies wholesale, or through your OEM.
> >
> >How is he right? I just proved that he was wrong.
>
> To quote:
> >> >If you buy directly from the distributor, I believe it's about the
same
> >> >price as you claim.
The $150 price, not the $500 price.
------------------------------
Reply-To: [EMAIL PROTECTED]
Subject: Re: Microsoft says Linux threatens innovation
From: [EMAIL PROTECTED] (Ed Allen)
Date: Mon, 26 Feb 2001 09:01:02 GMT
In article <Qrhm6.50$[EMAIL PROTECTED]>,
Erik Funkenbusch <[EMAIL PROTECTED]> wrote:
>"Ed Allen" <[EMAIL PROTECTED]> wrote in message
>news:[EMAIL PROTECTED]...
>> >>Software is not hardware, and it doesn't follow the same market trends
>of
>> >>hardware.
>> >
>> >Then why did you list only the most trivial hardware components above?
>> >
>> Could it be that M$ makes mice and keyboards ? I do not know about
>> cases and power supplies but I suspect that those had reached
>> commodity prices by many years of lots of radios and televisions
>> requiring them.
>
>MS doesn't have a monopoly on keyboards and mice. Logitech makes a
>significant number of them, so does Cherry (In fact, Cherry makes more
>keyboards than Logitech and MS combined).
>
Not being the sole provider of something is not what guarantees that
you have a monopoly.
It is being able to fix prices at above competitive levels.
Good evidence of that is sustaining profits of greater than 30% for
a number of years.
M$ still does at least twice that even with their recent woes.
--
How much do we need to pay you to screw Netscape?
- BILL GATES, to AOL in a 1996 meeting
------------------------------
From: "Erik Funkenbusch" <[EMAIL PROTECTED]>
Subject: Re: Microsoft says Linux threatens innovation
Date: Mon, 26 Feb 2001 03:18:58 -0600
"Chris Ahlstrom" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> Erik Funkenbusch wrote:
> >
> > > >>
> > > >> The price of Microsoft software has not stayed level. It has
increased
> > > >> by an incredible amount. For example, about 3-4 years ago, I
bought
> > > >> an OEM version of Office for about $150. The same version is now
about
> > > >> $500. How's that for monopoly practice!
> > > >
> > > >Not even.
> > > >
> > > >http://www.pricewatch.com/1/146/1904-2.htm
> > > >
> > > >Office 2000 Standard OEM costs roughly $175, and that's through a
> > middleman.
> > > >If you buy directly from the distributor, I believe it's about the
same
> > > >price as you claim.
> >
> > You can only buy OEM copies wholesale, or through your OEM.
> >
> > How is he right? I just proved that he was wrong.
>
> I was talking about Office Pro. On page 4 of 8 of the above link, I see
> a price of $255. That's still less than the price I quoted before,
though,
> which came from a PC Mall catalog.
How exactly did you quote an OEM price from a retail catalog?
> The price you quoted was for an upgrade, too.
No, there's no such thing as an OEM "upgrade".
OFFICE 2000 PROFESSIONAL OEM - SYSTEM RESELLERS ONLY - NO END USERS OEM -
full version $ 179
Office 2000 Professional Single User License (Resellers Only - No End Users)
$ 175
> If you go to page 6, Office Pro Retail Box is $420.
Office 2000 Pro contains much more software than Office 95 Pro did. Office
95 pro was MS Mail, Word, Excel, Access, Powerpoint. Office 2000 Pro
contains Outlook, Publisher, Word, Excel, Access, Powerpoint, and Small
Business Tools. That's Two very large extra applications (SB Tools, and
Publisher), and an extremely extended functionality application (Outlook).
If you were talking about Office 97 then it still didn't include Publisher
or the SB Tools.
Office 97 Pro included Access while Office 97 standard didn't IIRC.
> The last page shows BackOffice Small business server and Office Pro with
> 5 (count'em, 5) clients for $2045.
That includes Windows 2000 server with 5 clients.
------------------------------
From: [EMAIL PROTECTED] (Donovan Rebbechi)
Crossposted-To: alt.linux.sux,alt.destroy.microsoft
Subject: Re: Mircosoft Tax
Date: 26 Feb 2001 09:26:22 GMT
On Mon, 26 Feb 2001 00:47:28 -0700, Dave wrote:
>On Sun, 25 Feb 2001 19:40:43 -0600, "Erik Funkenbusch"
><[EMAIL PROTECTED]> wrote:
>
>Amazing. MS has achieved the impossible - they've exceeded the US
>Government in inefficiency! I mean, look at how much more Linus
>Torvalds and his merry band have accomplished for practically nothing,
>and here it takes MS "100's of millions or even multiple billions" to
>create merely incremental improvements to thier sorry excuse for an
>OS.
No, Linus and friends certainly haven't accomplished it for "practically
nothing". They are generous enough to volunteer their valuable time
towards Linux development. The difference is not really that Linux is
cheap -- the difference is that Linux is a gift. And a very valuable one.
--
Donovan Rebbechi * http://pegasus.rutgers.edu/~elflord/ *
elflord at panix dot com
------------------------------
From: "Mart van de Wege" <[EMAIL PROTECTED]>
Subject: Re: Mircosoft Tax
Date: Mon, 26 Feb 2001 10:34:39 +0100
Crossposted-To: alt.linux.sux,alt.destroy.microsoft
In article <1Mhm6.57$[EMAIL PROTECTED]>, "Erik Funkenbusch"
<[EMAIL PROTECTED]> wrote:
snip some stuff>
> So in other words, you agree that not all components have come down in
> price in recent years, and that the premise only applies to components
> which have a performance issue associated with them. The current model
> goes down in price when a better performing product replaces it.
>
> OS's generally do not improve their performance significantly from
> release to release. There is usually some, but the releases are years
> apart versus the months apart of hardware items.
<snip some more>
So, given that the performance of the underlying hardware does increase,
that would mean that OS's actually decrease in relative performance. I am
quite sure you didn't mean to say that, but that's the way your statement
parses. Care to clarify a little? Otherwise you'd be flamebait for all MS
haters who'll keep on harping that the various incarnations of Windows
keep getting slower for the same price.
Mart
Lost my sig :-(
------------------------------
From: "Edward Rosten" <[EMAIL PROTECTED]>
Subject: Re: M$ doing it again!
Date: Mon, 26 Feb 2001 09:57:58 +0000
> A "published" API is one which intended for external use. There are
> plenty of functions in the linux kernel which are not intended to be
> used outside of the kernel itself. Those are unpublished, and
> undocumented since they are not listed in man pages (unlike the
> functions which are intended to be used externally, which do exist in
> the man pages).
If you make up your own definition of published, then you are correct.
Using everyone elses definition you are wrong. All parts of the Linux
kernel are documented and published. Therefore there are no undocumented
and unpublished calls.
Since you are such an expert on the matters, why don't you go and name a
few `unpublished' calls, then.
-ed
--
| u98ejr
| @
Share, and enjoy. | eng.ox
| .ac.uk
------------------------------
From: SomeoneElse (SoneoneElse)
Crossposted-To: alt.destroy.microsoft
Subject: Re: Mircosoft Tax
Date: Mon, 26 Feb 2001 09:53:29 GMT
Reply-To: Truthteller
On Sun, 25 Feb 2001 21:15:05 -0800, Amphetamine Bob
<[EMAIL PROTECTED]> wrote:
>Seems this monopolizing crap has been going on for some time now.
>Just the stupid government never did much about it. Recall that
>Reagan was strongly against antitrust enforcement.
Yep he was. All those that think Macroshit is in the clear now because
Bush is in place should keep that in mind. Remember it was the Reagan
administration that broke up AT&T.
------------------------------
From: "Edward Rosten" <[EMAIL PROTECTED]>
Subject: Re: M$ doing it again!
Date: Mon, 26 Feb 2001 09:59:41 +0000
In article <yDom6.83$[EMAIL PROTECTED]>, "Erik Funkenbusch"
<[EMAIL PROTECTED]> wrote:
> "Aaron Kulkis" <[EMAIL PROTECTED]> wrote in message
> news:[EMAIL PROTECTED]...
>> Erik Funkenbusch wrote:
>> > All OS's have unpublished API's, including Linux. You can find all
> kinds of
>>
>> Wrong on two counts.
>
> Really?
>
>> > API's that are inside the kernel but do not appear in man pages
>> > because they're intended to be used internally inside the kernel
>> > itself without
> a
>> > published interface.
>>
>> 1) If it's only meant to be use INTERNAL to the kernal, and
>> *NEVER* by application, then it's *NOT* part of the API.
>>
>> API = APPPLICATIONS Programming Interface.
>>
>> Anything which is "internal to the kernal" is, by definition, outside
>> of the scope of the API.
>
> That's an awfully narrow view? What exactly do you call the kernel
> module API? A kernel module is not an application, yet the kernel
> exposes a set of interfaces for the modules to link to. Those
> interfaces are commonly known as API's, even if it's not an application.
Pedantry is neither clever nor funny.
>> 2) What part of OPEN SOURCE do you not understand?
>>
>> The source-code IS *ALWAYS* the authoritative documentation.
>
> Not everything in the kernel is documented in source comments either.
Then read the source code. It's all in there.
-Ed
--
| u98ejr
| @
Share, and enjoy. | eng.ox
| .ac.uk
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list by posting to comp.os.linux.advocacy.
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Advocacy Digest
******************************
