Letters to the editor

Only a fool would believe a Microsoft study after this!

This Microsoft-sponsored Microsoft-engineered report caught my eye, but 
a few things in it really were outstanding, and I mean outstanding like 
large lime green and orange paint squares chequerboarded onto a Rolls.

    http://www.veritest.com/clients/reports/microsoft/mssmbiz.pdf

The "Linux consultants" mentioned in this study wanted 24 programmer 
hours to be able to automatically email out server stats?

    unlike Windows SBS 2003, the monitoring and reporting solution
    used for Red Hat Enterprise Linux ES did not support sending
    performance and usage reports to an email address at regular
    intervals. Linux consultants estimated that this support could
    be added through custom scripting at a cost of 24 hours of
    development time.

urpmi wget metamail
cat > send-usage-updates.sh
    #!/bin/sh
    #
    # send usage charts by email
    # works OOtB for MRTG, RRDtool and Webalizer but for the URL
    # might want to change to and from addresses to suit
    #
    SCRATCH=/tmp/stats-$!.tmp
    mkdir $SCRATCH
    cd $SCRATCH
    wget -nH -nd -p http://url.of.stats/page/index.html
    metasend -F tux-AT-propaganda.morons.inc -z -t billg-AT-morons.inc \
      -s "Stats for $(hostname) at $(date)" -e base64 -b \
      -D "Stats for $(hostname) at $(date)" -f index.html -m text/html \
      $(ls *.png | gawk '{ print "-n -m image/x-png -D image -f",$1 }'
    cd
    rm -rf $SCRATCH

Seventeen minutes including testing ("yup, them's my server stats"), 
while doing other stuff in background. Gawrsh, that was hard. AUD$34.00 
at my normal rates, except I have a one-hour minimum.

I guess I have to charge AUD$120.00*24*60/17 == AUD$10165.00 an hour for 
programming time now, to stay on the same efficiency level as 
Microsoft's amazing Linux consultants.

If our performance ratios are generalisable, it would have taken me two 
minutes and three steps to do the Linux OEM install, three minutes and 
four steps for the full install (roughly 29 and 75 times faster than 
Microsoft SBS 2003, respectively).

I'm not quite that good, just ask anyone who knows me, but experience 
tells me that whacking in an enterprise edition of Mandrake 9.1 as far 
as Step 1 with Yes against every single feature listed for both 
platforms, plus a whole lot more, takes about 35 minutes for a single 
internet domain on a dual P3-1000 box with 2GB of RAM and paired 18GB 
SCSI hard drives.

With Wayne's permission, I can show you that box running thin clients 
today. This is not a theory.

I'm not a Red Hat fan, but I don't understand how Microsoft's testers 
managed to *avoid* Red Hat's installation wizard - maybe it didn't have 
"Wizard" printed all over it, or had more than one choice per page? 
Shrug.

Again I can't speak for Red Hat, but setting up a DHCP server on 
Mandrake is one checkbox in DrakConnect. If ("Ooh, Lordy, Lordy! Oh, 
please, Brer Fox, don't make me edit the config file!") I had to edit a 
DHCP config, it's all of - what, six or eight lines of code? Horrors!

I feel compelled to ask this: is a person incapable of doing or 
unwilling to do that very simple chore the kind of person you want 
running your Internet-exposed servers? Really?

As for wizards reducing download/install times for new packages, even 
Red Hat's very boring GUI package manager must have been too awesome 
for these skilled testers to dirty by touching. It sounds kind of like 
they were drag-racing against a sleigh in summer.

I must admit, however, that Windows SBS 2003 does win on Step 3, "Build 
an intranet web site for information worker collaboration". Sounds 
impressive, doesn't it? Um, question for VeriTest: how do zero steps 
occupy seven minutes and nineteen seconds?

Putting up a Wiki or PostNuke does involve either opening a shell and 
typing a short one-liner or 5 clicks to accomplish (either 29 or 17 
times more efficient than their "Linux consultants").

Again I'm pressed to ask: isn't Windows constantly getting into hot 
water for having services switched on by default? Why then is it a good 
thing that SBS 2003 arrives with them enabled?

I'm also wondering how "urpmi vncviewer rdesktop openssh-server" and the 
installation of PuTTY and TightVNC on the XP workstation managed to 
chew up over 100 minutes for VeriTest's Linux gurus. That's all you 
need to do to complete Step 4 on Mandrake, maybe twenty seconds for the 
server and three or four minutes on the XP workstation for full 
connectivity both ways.

You could even add vnc2swf to that urpmi line and make Flash movies of 
the XP box doing stuff while you waited for the Microsoft guys to win 
the battle with their wizards.

Maybe I should apply for a job at VeriTest? I could charge triple time 
and they'd still save buckets of money on consultants.

It seems the war of words has degenerated here into a war of headlines. 
Anyone looking at the details with half a brain will be either totally 
gobsmacked or laughing too hard to protest. It seems that either 
VeriTest have shifted their research labs into the Ministry of Truth 
building - and no longer seriously claim objectivity - or their 
calendar is six months out of sync.

This only serves to throw Forrester's recent decision to actively avoid 
participating in such charades into sharper relief. It's nice to see 
that at least a few consultancies still take their audience seriously.

Cheers; Leon

-- 
http://cyberknights.com.au/     Modern tools; traditional dedication
http://plug.linux.org.au/       Committee Member, Perth Linux User Group
http://slpwa.asn.au/            Committee Member, Linux Professionals WA
http://linux.org.au/            Committee Member, Linux Australia

Comments (1 posted)

After Sun goes out

Sun Microsystems crossed the line from "troubled" to "doomed" yesterday.
This is sad news for the open-source community, and we need to think
about how we're going to deal with it.  The most pressing questions
are "What becomes of Java?" and "What becomes of OpenOffice.org?"
These are questions that matter.

Sun's troubles have been mounting for a while.  Founder Bill Joy's
departure was an ominous recent symbol, but the substance of their
problem is that their hugh-margin server business is being eroded from
the low end by PCs running Linux at a rate that doesn't leave it
much of a future.

Nobody should cheer the prospect of Sun's demise.  Sun screwed up some
major decisions very badly, from wrecking Unix standardization efforts
in the 1980s to throttling the dream of Java ubiquity by keeping the
language proprietary.  But nobody should forget that Sun was founded
by Unix hackers for Unix hackers.  For most of its lifespan Sun
remained the archetype of an engineering-driven company. Sun was,
mostly, among the good guys; to hackers and geeks, disputing with Sun
was almost a family quarrel.

But inside Sun, I hear that talent is bailing out of the company
because they just don't believe the Solaris-will-prevail story
management is peddling.  Most of Sun's techies are running Linux on
their PCs at home.  They can see the handwriting on the wall.

In retrospect, the recent pronunciamento that Sun has no Linux
strategy was their final admission of failure. Sun can't run at the
lean profit margins that are all a commoditized Linux server market
will support, their cost structure is all wrong for it.  They got
trapped in a classic innovator's dilemma and didn't cannibalize their
own business while they had the investor confidence and maneuvering
room to do so.  Cuddling up to SCO didn't help, either.

And now it's too late[1]. Moody's has just about dropped Sun into the
junk-bond basement.  The stock closed at $3.31, 15% off for the day
and falling in heavy trading.  The recent product announcements have
been duds, and the upcoming quarterlies are going to be a disaster.
Wall street analysts are calling for drastic job cuts and speaking the
code phrases that mean "run for the hills!"  The smell of death is in
the air.

Any of Sun's people and tangible assets that don't scatter to the four
winds will probably wind up in the hands of IBM, HP, and Dell -- three
companies that have shown they do know how to play the
commodity-computing game.  The SCO lawsuit probably won't be
affected. Sun was the lesser-known of of SCO's sugar daddies along
with Microsoft, but Redmond can pick up Sun's share of funding the
lawsuit out of petty cash -- and it undoubtedly will.

The real question is twofold: can OpenOffice.org survive without Sun, and
where will Java land?  Probably not at Microsoft; with C# in the 
picture, it is unlikely that Microsoft even wants to own Java any more.
I have to guess that IBM is the most likely to shoulder both technologies, 
simply because nobody else is really positioned to do it.  But that,
of course, raises other worries -- is it really good for us if IBM
has a lead position in everything?

[1] http://reuters.com/financeNewsArticle.jhtml?type=hotStocksNews&storyID=3535714
-- 
		<a href="http://www.catb.org/~esr/">Eric S. Raymond</a>

All governments are more or less combinations against the
people. . .and as rulers have no more virtue than the ruled. . .
the power of government can only be kept within its constituted
bounds by the display of a power equal to itself, the collected
sentiment of the people.
	-- Benjamin Franklin Bache, in a Phildelphia Aurora editorial 1794

Comments (none posted)

Page editor: Jonathan Corbet