Linux-Advocacy Digest #824, Volume #31           Mon, 29 Jan 01 16:13:06 EST

Contents:
  Re: How long does your box run for? (R.E.Ballard ( Rex Ballard ))
  Re: Desktop MTTF, Linux, lets get some numbers. (Bruce Scott TOK)
  Re: Microsoft DEATH NECKLESS is COMPLETE!!! (Peter Hayes)

----------------------------------------------------------------------------

From: R.E.Ballard ( Rex Ballard ) <[EMAIL PROTECTED]>
Subject: Re: How long does your box run for?
Date: Mon, 29 Jan 2001 20:33:12 GMT

In article <[EMAIL PROTECTED]>,
  Michael Vester <[EMAIL PROTECTED]> wrote:
> Matthias Warkus wrote:
> > =
>
> > It was the Sat, 27 Jan 2001 11:30:15 -0700...
> > ...and Michael Vester <[EMAIL PROTECTED]> wrote:
> > > 144 days of uptime the longest my Linux boxes have. The only
> > > reason they are restarted is because like California, here in
> > > Alberta we are going through a power deregulation too. Power
> > > deregulation is the stupidest thing any government has ever
> > > done.

Actually, there was an even bigger problem.  While the federal government
deregulated the utilities under the Pressler Act, and the east coast
deregulated their utilities under companion legislation, the west coast and
southwest didn't deregulate.  As a result, state utilities weren't allowed to
sell excess capacity at competitive rates and weren't allowed to pass on the
short-term costs of initial contracts in exchange for long-term reductions.

The Silicon Valley region was especially vulnerable since they depend heavily
on wind-mill generated power from windmills just west of Stockton CA and
running along that ridge of mountains.  When the wind blows strong, there's
more than enough power.  When the wind is calm, the power has to come from
somewhere else, and it must be purchased at the competitive rates.

> > Hey, we in Germany deregulated power
> > two years ago or such, and still
> > our power grid is about five times
> > as reliable as the US one used to
> > be. Deregulation is OK, you've just
> > got to do it the right way.

The problem in the U.S. is that there are numerous regulatory
agencies in each state, each of which have different vested
interests.  The federal deregulation made it possible for
companies who had excess capacity to sell that capacity to
the highest bidder.

The problem in California was that the PUC was still restricting the power
company's ability to purchase at competitive rates.  Since the utilities
couldn't pass on the cost of extended contracts at higher than maximum prices
established under the regulated economy, the utilities could only purchase
power on the "spot market" which meant they had to pay even more to meet
demand (which they still couldn't pass on to rate-payers).

Eventually, with no way to generate excess power, consumer states such as
California, who purchased exess capacity from Colorado, Utah, and Nevada at
federally regulated rates suddenly found themselves competing with New York,
Chicago, and Dallas who were also deregulated and were thus able to purchase
extra capaacity at higher rates that were lower than the "spot market" rates.

California rate payers limited to 10 cents a KWH while New York
and New Jersey were willing to pay 12 cents a KWH.

> We knew that we would not have enough production capacity.
> Deregulation works if there is some extra capacity.

Nationwide, there was plenty of extra capacity, but not at
the rates California was legally allowed to pay.  The utility
companies did their best to buy extra capacity even though
they were losing money.  Unfortunately, the spot rate was nearly
double the rate that the utilities were allowed to charge
California residents.

Had California deregulated at the same time as everybody else, they might
have had a short-term increas of 10%, and there would have been incentive for
California to reactivate some of their idle capacity so that they could
become a producer instead of a consumer. Instead, with regulations that
prevent the power plants from selling their excess and prevent them from
raising rates to purchase from other grids at competitive rates, they have
been squeezed to the brink of bankruptcy.  Now, the state government will
have to either deregulate quickly enough to allow the local company to
recover, will have to "bail out" the company, or will lose it's local
utilities to an interstate utility that will push for even more deregulation.

The reality is that in a competitive environment, there will be more than
enough power to go around.  Furthermore, consumers can choose their preferred
power source.  They can choose from coal plants that emit green-house gases,
neuclear plants that may emit low level radiation and will require effective
fuel management, or solar and wind power which may require frequent purchases
from the other sources at higher rates due to changes in the weather.

The power producers will eagerly provide exess capacity if there are
companies willing to commit to long-term contracts that will provide a stable
revenue base at profitable rates.

> But still, my computers are running Linux and the only thing
> that causes any downtime is the supply of electricity. Since
> my computers are not running anything that lives depend on,
> downtime is no big deal. I still have far less downtime with
> Linux and poorly implemented power deregulation than I had
> using Windos with steady, reliable electricity. =

I finally broke down and purchased an APC last year.  The battery
lasts 30 minutes (since I don't plug the monitor into the APC.
I have 3 PCs plugged in and the Linux machines can stay up
for months at a time.

Today, Netcraft now offers "uptime" listings for various
sites, including a breakdown of uptimes by server type.
Windows NT sites seem to last about 7 days.  Windows 2000
sites seem to last about 20 days, Linux seems to last around
3 months, and Solaris lasts about 6 months.

Note that this is only time between reboots and includes both
scheduled and unscheduled outages.

> I have spent over two years mostly lurking in this group. I
> have grown fond of the many interesting characters. Flatfish
> or whatever you call yourself, keep up the effort! If you are
> the best that the pro-MS side has, then Linux has already
> won.

>  Speaking of characters, what happened to Dresden Black?

Drestin Black has been very quiet lately.  He posted a number
of proud announcements about how Hotmail was being converted
to Windows 2000.  In reality, only the web server component
was being switched over and the results were dismal.

Microsoft purchased hotmail because they wanted to prove the
superiourity of Windows 2000/Back-office to FreeBSD and Solaris.
This far, the core back-end functionality is still FreeBSD
and Solaris.

> He managed to always have an URL to a MS-infomercial when
> expressing the virtues of NT.

Drestin may have joined the ranks of the Linux community.  Many
of his concerns and frustrations with Linux have been resolved.
Mandrake 7.2 solved many of the installation problems with
traditional hardware and the Linux 2.4 kernel is resolving many
of the USB problems.

> I hope he is happy changing
> toner cartridges and re-installing Office.  I look forward to
> the wisdom and experience of Rex Ballard postings.  Rex is my
> Linux hero. He is both credible and sincere.

I'm glad you enjoy.  I'm compiling some of my posts which should be
up in a few days.

> And thank you
> Aaron R. Kulkis for the two or three extra keystrokes to read
> your posts.  Love your message, hate your sig.

Same here.  I usually snip all but my favorite part.

> Still have to use W2K at work. Still freezes once or twice a
> day. It is much better than NT4 which froze 2-4 times a day.

I've been finding that many of the Windows applications now
seem to be running quite well under Wine.  VMWare is a bit
slow on my 200 Mhz Pentium laptop (which is fine for Linux).

> I develop database applications using Oracle,
> MySQL, Access on Solaris or NT.



> The customer dictates the platform that I use. Our
> Solaris server currently hosts over 900 dynamic web sites
> using Netscape web server.  The IIS servers peak out with 5
> customers max with dynamic web sites.  The Solaris server is
> never down. The IIS servers must be regularly rebooted or face
> the expected but unwelcome freezing.  Customers with no OS
> preference are always steered towards the Solaris.  We are
> moving from Netscape to Apache this year and will try to port
> some of the IIS stuff over too.

There is now ASP support for Apache.  I haven't really torture
tested it, but it seems to do a reasonably good job at ASP style
applications - and it runs on Linux.  You can also use progreSQL
or mySQL as well as Oracle, Sybase, and DB2.

> Customers notice really fast
> when their sites are dead in the water. IIS has proven to be a
> very flimsy base to build dynamic web applications on. =

Just out of curiousity, have you considered offering Linux or
FreeBSD as an alternative?  Some people want a more "private"
host without the extra costs of purchasing an entire server.

Ironically, Windows NT and Windows 2000 servers tend to get support
primarily from

> Upgrading to W2K server has proven to be a daunting task. The
> MCSE's hare at a complete loss on how to do it. My suggestion
> is to build a W2K site and put new web sites on it. The NT4
> servers will have nothing new developed on them. Very
> wasteful. =

So is developing new applications on Windows 2000.  Windows 2000
has definite advantages over Windows 2000, but still pales in
camparison to the capabilities of Linux.  In fact, it's hard
to keep track of all the new features of Windows 2000.

> Most places I have worked for,  middle management have been
> brainwashed with Microsoft propaganda for a long time. The
> lure of low TOC was too much to resist.

Unfortunately, Middle management is often lured by a combination of factors,
including familiarity with Windows NT Workstation (which makes them think
they know something about server design), the lure of fame and fortune by
being associated with "Billy's Billions", and Microsoft's willingness to
provide billions in advertizing in the magazines that middle managers with 
an eye on senior management tend to read (Wall Street Journal, Barrons,
Business Week, Fortune).

One of the biggest breakthroughs for Linux was when Forbes magazine
published an article that covered the leaders of the Open Source
movement.  In fact, when Forbes asked for the most notable people
of the 20th century, Linus Torvalds, Richard Stallman, and Tim
Berners-Lee scored as prominantly as Bill Gates and Steve Jobs.

In fact, many of those who cited the accomplishements of Bill Gates
were actually citing the innovations and accomplishements of Torvalds,
Stallman, Lee, and Bill Joy.

>  Now, flaky computer software is accepted.

Actually, corporations are now finding more and more resistance
to Windows Servers.  In fact, there are many companies who have
reached the point of hiring "targets".  They hire someone who is
knowledgable and ask him to "sign off" on a Windows 2000
reccomendation so that if Windows 2000 blows up, he, or she,
will be the one fired rather than the executives who are pushing
for the decision.  It's a suicide job, a bit like being a Kamakazee.
You are paid well and treated well until the day comes that they
need to fix the blame on you.  Then you have to defend your decision
before they dismiss you.  There are some managers who actually
specialize in being "suicide kings".  They believe in Windows 2000
so strongly that they reccommend it even though it blew itself to
kingdom come in the previous assignment.

> The server going down is an anticipated ritual.

This is the primary reason why the "Mith of lower Microsoft TCO" is so
amusing.  Ironically, many people try to blame the Intel platform rather than
the Operating System.  What is interesting to note is the number of Linux
systems running on Intel hardware that runs flawlessly for months, even
years, that is inferior to the hardware used for Windows NT and Windows 2000.

Ironically, Linux has become so popular as a server that there is a whole new
market in rack-mounted Linux machines based on Intes hardware.  In same
cases, there isn't even a connector for a video monitor since the system
boots

> Saving your work every 10 minutes becomes a nessecary
> habit.  Like the story of the emperor with no clothes, nobody
> will admit they don't really need another 1000 features in
> Word.

In fact, Microsoft has introduced a number of features that
customers really hate.  The "Dancing Paper Clip" that insists
on popping up when you are really in a hurry to get a report
finished is a really "popular" feature.

The myth of low Microsoft TCO is based an comparisons between a combination
of Sun servers and Netware servers combined to provide the equivalent
functionality of an NT server.  Of course this same report also suggested
that you would be able to serve 1000 users with only 3 Windows NT servers and
provide the same functionality as that offered by Solaris and Netware.

What they didn't mention is that You could run SAMBA on Solaris
which meant you didn't NEED Netware.  Furthermore, the cost
of a Solaris server with equivalent capabilities to the NT server
was actually less expensive.  Since a monoprocessor Solaris machine
could outperform a quad processor NT machine, you got more bang
for the buck with Solaris.

The original estimate of 3 servers per thousand users was, to put it
politely, optomistic.  In reality, you needed nearly 40 quad-processor
servers per thousand users compared to a single quad-processor UltraSparc
server per thousand under Solaris. That's 160 processors compared to 4.

To make matters worse, Windows was vulnerable to memory corruption
which made quad processors and RAID arrays ineffective at preventing
server failures.

When you actually compare Windows NT or Windows 2000 to Linux,
Linux completely blows the bottom out of the TCO picture.
As a server, Linux goes nearly 10 times longer between reboots
(scheduled or otherwise), Linux make more efficient use of Memory
which means that you can serve more users with fewer servers.

Linux can serve over 1000 users on a single Netfinity server, and
provides multiple features under that server.  Furthermore, the
superior memory management of Linux reduced the risk of failures
due to memory corruption.

Many CEOs, CFO,s and CIOs had no clue that the backbone of their information
systems were actually based on Linux technology.  Often, the outlook
back-office server is replaced with Linux and the outlook users don't even
know it's been switched, they just notice that the mail server stopped
crashing.  They notice that their file-server works better, and their
intranet web sites seem more flexible.

Many times, the boss will start to specify a system and a Linux
prototype is generated before the server arrives, because the team
was able do dig an old "boat anchor" out of a broom closet and use
it as a prototype server.

There are even consulting firms who sell a package that essentially consists
of Open Source applications, some proprietary packages such as SAP, MQSeries,
and WebSphere or WebLogic coupled with an infrastructure of Open Source
"glue" to customize product to customer needs.

Linux also make a good database server for smaller databases, and
Linux 2.4 makes a good server for multiterrabyte databases.

When you do finally out grow Linux (getting harder and harder
each month) you can easily migrate that Open Source code to
Solaris, AIX, or even Z-900 servers (mainframes).

> With an aggressive marketing campaign, Microsoft are trying to
> convice everyone that they are now producing "enterprise"
> quality software.

Microsoft's advertizing budget is formidable.  You can see television
ads everywhere.  It's their last defense against the rapidly evolving
Linux platform and UNIX systems.  Sun wastes a lot of money with it's
gimmicky commercials.  They really need to drive home the point that
85% of the time, users are using UNIX.

IBM does a really nice job of presenting their message.  The
administrator's nightmere, the server vendor blames the software
who blames the database who blames the the network who blames the
application who blames the server.  And who's responsible?  The
guy who thought Microsoft was "the solution".

>  Their tv commercial shows a picture of a
> server in a little room, claiming
> to be running the comapany's
> new web site. They claim that no one
> has "visited" the server
> in days. Days??? I guess that is good for Microsoft.

Yes.  Microsoft now has an enhanced remote management
capability similar to SMS.  They still have to come in
to reboot the machines, and if they loose control of
the console, someone has to pay them a visit.

What would be really funny is opening up some of the
closets that are still running Red Hat 4.0 (Cobwebs,
blow off the dust) the guy in the pin-stripe suit
says "What's that",

the guy in the golf shirt says;
 "That's our e-mail server, it's been running since 1996
  without a reboot".
The suit says; "I thought that was our e-mail server"
   (pointing to the Windows machine).
The golf shirt sheepishly says; "well, not really".
The suit says; "what do we use that for".
Golf shirt says; "Video Games".

> Our
> Solaris server is 200  miles aways and I don't think it gets
> any visitors in months. Congratulations Microsoft if you can
> keep a W2K server running for days without the intervention of
> a MCSE. =

And of course, there is the SCO (now Caldera) machines that
sit in the closets of Burger Kings, Taco Bells, Kentucky
Fried Chicken, and others.

> Where I work, we depend on the efforts of our MCSE's to keep
> our paltry 50 dynamic websites runnning on 10 servers. Also,
> they do a great job replacing toner cartridges and
> reinstalling applications if they get enough time.  Those 10
> servers keep them hopping. It takes only one Unix admin
> working part time to maintain the Solaris. Lets see, 3
> fulltime MCSE's to service 50 dynamic websites or one part
> time Unix admin to service 900 dynamic web sites.  Am I wrong
> or is our TOC higher for the MS platform?  =
>
> Microsoft Total Cost of Ownership (TCO) =  Total Cost of
> Crashing (TCC) + Total Cost To Switch to Something That Works
> (TCTSTSTW)

PLUS the total cost of obligatory upgrades (TCOU), total cost of
hardware upgrades (TCHU).

Keep in mind TCC = Total Cost of Staff who reBoots box (TCSBB) +
   Total Cost of Staff who reinstalls boxes" (TCSRB) +
   Total Cost of Time lost to rebooting boxes (TCTLRB) +
   Total Cost of Time due to lost or corrupted data (TCTLCD).

Put simply, switching from Windows to Linux over a 4 year period
could double productivity every 9-12 months.

> Michael Vester
> A credible Linux advocate.
>
> > mawa
> > --
> > Einblattlocher!
> > Elektrogriller!
> > Erbsenp=FCrierer!
> > Zweifingertipper!
>

--
Rex Ballard - Sr I/T Systems Architect
Linux Advocate, Internet Pioneer
http://www.open4success.com
Linux - 80 million satisfied users worldwide
and growing at over 9%/month! (recalibrated 01/14/00)


Sent via Deja.com
http://www.deja.com/

------------------------------

From: [EMAIL PROTECTED] (Bruce Scott TOK)
Subject: Re: Desktop MTTF, Linux, lets get some numbers.
Date: 29 Jan 2001 21:40:22 +0100

In article <9Wid6.15$[EMAIL PROTECTED]>,
Erik Funkenbusch <[EMAIL PROTECTED]> wrote:
>Self-selected surveys are nowhere near accurate.  In general, only the
>people with good results are going to send you data, despite the obvious
>weigthing of the survey by your parameters.

You have experience with Microsoft surveys, I see...

-- 
cu,
Bruce

drift wave turbulence:  http://www.rzg.mpg.de/~bds/

------------------------------

From: Peter Hayes <[EMAIL PROTECTED]>
Crossposted-To: comp.os.ms-windows.advocacy,comp.os.ms-windows.nt.advocacy
Subject: Re: Microsoft DEATH NECKLESS is COMPLETE!!!
Date: Mon, 29 Jan 2001 20:57:38 +0000
Reply-To: [EMAIL PROTECTED]

On Mon, 29 Jan 2001 04:02:20 +0200, "Ayende Rahien" <[EMAIL PROTECTED]>
wrote:


> > Posted from OE 5.6 using Whistler AD 2296, Uptime 17 hours 33 minutes 51
> > seconds, 1 user @ console, 3 users via TS, 2 diconneted user sessions.
> > The low uptime is just because I'd to just install the system.
>                                                     ^^^^^^
> 
> Typo, should say, I'd just installed the system.

As opposed to "just re-installed the system" as per the usual Microsoft fix
for anything that goes wrong???

Only kidding...... or am I?

Peter

------------------------------


** 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
******************************

Reply via email to