See also: last week's Letters page.

Letters to the editor

From:	 Con Zymaris <conz@cyber.com.au>
To:	 letters@lwn.net
Subject: Re: Why is the support business so hard?
Date:	 Tue, 8 May 2001 14:34:51 +1000


LWN asks:

> Linux has taken off, and the support options exist.  So why are so few
> companies buying those support services?  Perhaps there are far fewer
> important Linux deployments than people think.  Without deployments,
> there is little need for support contracts.  We don't believe it, though

You ask the right questions, now I'll happily provide our version of some
answers. In short, Linux is causing a small boom in our systems
professional services business in Australia. The market is there if you
want to work it.

First, some background. Cybersource has been successfully providing
Unix/Linux/Internet Professional Services in Australia for 10 years. Linux
has gone from being a small part of our revenues, to perhaps the largest
part, in the space of the last 4 years. Our target market is broad. SMEs,
Government and Corporate. While it's true that for the most part, the
majority of the growth in Linux services has been in the SME area, this is
changing. 

Perhaps the big-name US-based support organisations who have been
experiencing problems have been trying to pitch business primarily to the
larger customers; these same customers who are only now moving into Linux.
Due to the cost of overheads (very high-profile advertising, largish
instant staff, expensive high-profile location offices) that some of these
big-name Linux support organisations carry, they actually _need_ to target
customers in the higher margin corporate and government. It is our belief
that to start small (Cybersource has only 40 staff) and grow organically
through word-of-mouth, befits the Linux/Open Source market better, than to
start with a big-expenditure splash, as made in recent years by the
various big-name Linux support start-ups. Grow with the market, not ahead
of it.

In short, the demand is really out there. Join us in bringing Linux and
free software to the business world.


Cheers,

Con Zymaris
CEO
Cybersource   

 -- 
_____________________________________________________________________________
Con Zymaris <conz@cyber.com.au> Level 9, 140 Queen St, Melbourne.  9642 5997
Cybersource: Successfully Providing IT Professional Services for 10 Years
Specialists in Unix/Linux, TCP/IP and Web App. Development  www.cyber.com.au

From:	 "CARNIELLO, MIKE L. [FIN/1820]" <mike.l.carniello@pharmacia.com>
To:	 "'letters@lwn.net'" <letters@lwn.net>
Subject: Advocacy, not unreasonableness
Date:	 Thu, 3 May 2001 14:16:45 -0500

To the Editor,

Your recent comments regarding Linuxcare (03-MAY-2001) indicate that perhaps
it's time to yet again adjust your rose-colored glasses you seemingly use
for OpenSource/Linux issues. You mention:

"What if the truth were something else: what if Linux users simply do not
need support? ... Could it be that, in the end, technical support services
are only needed for proprietary, black-box systems?"

Oh, come on! Linux is incredibly complicated operating system to use and
maintain, whether server-based or desktop-based. Support is needed for all
types who come in touch with a Linux system - end users, application
adminis, system admins, and hardware people. This support may be provided by
intra-company or external sources, but it still must be provided.

You go on to appropriate the corporate catchphrase 'empower' in writing:
"Free software empowers its users to take responsibility for keeping their
own systems going."

Empowers??? I think the word you're looking for is "forces." And that's not
necessarily a bad thing, but it is a double-edged sword.

Mike Carniello
mlcarn1@home.com

From:	 "Michael Farnbach" <mfarnbach@conneq.com>
To:	 <editor@lwn.net>
Subject: Support for Linux
Date:	 Thu, 3 May 2001 11:22:08 -0700

First, I have always loved your journalistic style.  But maybe the tone
on the front page of this weeks issue was a little too appologetic?

Either way I'd like to add my two cents being somewhat in the support
industry myself.  I remember calling Eklektix a while ago when you were
one of the only games in town when it came to Linux support, Liz truely
is cool.  Since then I have installed various machines in small
buisnesses and I can attest that they just run.  Our longest out box
just recently was brought in for service.  We updated it, added a raid1
and a journaling filesystem and a better web admin tool (we were using
swat and linuxconf).

The amazing part is that we hadn't touched, rebooted, been contacted by
them in the 18 months since we deployed it.  It just worked, and Time
flew by.  And since the client's office is pretty low on Linux knowledge
I can assure you they weren't kind to it and shouldn't be accused of
pampering or administring it themselves.  We haven't ever been called
for support on any of our other deployed boxes either.

Linux seems to be the perfect Drop and Forget server deployment tool for
a small IT outsourcing buisness like ours.

From:	 Rob Landley <rlandley@austin.rr.com>
To:	 letters@lwn.net
Subject: LinuxCare's "support" business.
Date:	 Fri, 04 May 2001 17:59:52 -0500

Making money from Linux tech support runs into two problems.  First of
all you don't need it, and secondly you can do it yourself.

First, most of the support people need is the "getting it to work in the
first place" variety.  Install and configuration is a one-shot deal, not
an ongoing revenue stream.  Once you've configured a reliable system, it
can get buried behind sheetrock during remodeling and nobody's likely to
notice for about five years.  (This has actually happened to novell
servers and PDP 8 systems. 
http://www.techweb.com/wire/story/TWB20010409S0012 ).  Perhaps you
contract out the installation of the system and take out an insurance
policy against anything going wrong the first few months, but after a
while there's no reason to keep paying for babysitting.

Secondly, if you're not going to totally outsource your information
technology infrastructure (not just a "tech support" contract but having
the servers and their caretakers live in an IBM data center), then
you're going to have an IT staff.  Even if it's just one guy, he'll have
the complete source code to everything and will be quite capable of
fixing things himself.  Perhaps he'll have to search a few newsgroups to
find the information he needs, but keeping it running will be part of
his job.

So LinuxCare's problem is that it either does too much or doesn't do
enough.  Red Hat provides install time support, and IBM provides
throw-money-at-the-problem complete solutions.  In between, there's just
not much revenue.

Linux has never been something you make money ON.  It's something you
make money WITH.

Rob

From:	 Derek Kite <derekkite@netidea.com>
To:	 letters@lwn.net
Subject: Support business so hard?
Date:	 Sat, 5 May 2001 20:00:38 -0700

Support in any industry is a treacherous business. I work in the 
refrigeration service industry, and the number of company failures are very 
high. The difficulties are due to the high level of competence required from 
not the managers or salesmen, but the people with the dirty fingernails. Good 
technicians are rare and rather independant minded, more likely to start 
their own small service company than work for a large firm, or would rather 
be part of a small organisation. The only advantage that a large firm has is 
connections to head office, and a depth of expertise that the likes of 
IBM. Otherwise, the only difference is a larger overhead.

Why would someone hire Linuxcare over the local small firm of competent linux 
technicians? I hope for their sake the reasons are clear in their customer's 
mind. All I know is that there will be many failures, especially of large 
firms that sell services. But there will be (and is) a large industry of 
small firms that will do increasingly well as linux becomes a common option.

Derek Kite

From:	 "John Carter" <john.carter@tait.co.nz>
To:	 <letters@lwn.net>
Subject: Package mechanisms break Open Source.
Date:	 Mon, 7 May 2001 11:28:16 +1200 (NZST)

Current distributions and package mechanisms break the power of Open
Source.

In the bad old days if you wanted a program you downloaded the source,
compiled and ran. If it died you fired up gdb, sniffed around, fixed it
and sent the patch in. If it lacked, you added code until it did what you
want. If you didn't know how things worked, you "Used the Source Luke".

Distributions and package mechanisms and the need to squeeze onto small
disk drives have removed the current generation from that.

Now disk drives have grown huge.

Distribution and Package tools should now by default put unstripped
binaries _and_ the source onto your drive. If a process segfaults, it
should drop you into gdb.

I'm willing to bet you the pace of Open Source evolution will increase by
a factor of a 100 if this recommendation is followed.



John Carter                             Phone : (64)(3) 358 6639
Tait Electronics                        Fax   : (64)(3) 359 4632
PO Box 1645 Christchurch                Email : john.carter@tait.co.nz
New Zealand

From:	 "james c" <james_dasfleet@hotmail.com>
To:	 letters@lwn.net
Subject: Someone To Sue
Date:	 Fri, 04 May 2001 15:56:19 -0000

I had to laugh when I read your item which quoted 32BitsOnLine as saying "I 
would sleep better knowing that I could shift blame to Bill Gates."

Does 32BitsOnLine think Mr Bill cares?

I've heard similar statements many times in my consulting career, usually 
from a manager who says something like "we have to buy commercial products 
so there is someone to sue if it goes wrong".

My usual response is along the lines of "So imagine we buy a database from a 
multi-national corporation, and something in it breaks and we lose a million 
dollars. Do you really think you can sue AcmeMegacorp/Microsoft/whoever? 
Their lawyers would take you apart, haven't you ever actually read a licence 
agreement?"

I'd much rather have a product with good support, or the source code so I 
can support it in-house, than one with the supposedly sleep-inducing 
properties of an un-sue-able megacorp behind it.

Cheers,
James

From:	 Max.Hyre@cardiopulmonarycorp.com
To:	 letters@lwn.net
Subject: Free-Software's impetus, contra Mr. Mundie
Date:	 Fri, 4 May 2001 15:43:01 -0400

   Dear LWN:

   Though it is true that repeated sales of Free Software is not a
viable business model, this observation only applies to that class
of people involved in making money by selling the software.  It
completely ignores the class of people making money using such
software as a tool.

   For this second class, the cost of software is a loss, mitigated by
its utility.  Getting that utility at a fraction of the cost will be
an extremely attractive proposition.  It makes sense for them to band
together with others, even competitors, to develop and improve
programs which are part of their infrastructure.

   Witness the Apache Group, which grew out of a number of webmasters,
for whom the server is a means, not an end.  Even if some of them were
business competitors, so long as that business wasn't selling Web
servers, they were better off cooperating to sharpen the tool.

   Such cooperation doesn't arise out of nothing.  But all it takes is
one generous soul to free a useful program.  That early, probably
minimal and buggy, program then serves as a focus about which the
larger group organizes.  Think of it as the impurity which starts
crystallization of a supersaturated solution.  The effects are all out
of proportion to the initial stimulus, but rather reflect the size of
the group which can fruitfully use the program.

   =That= is why a model that's unworkable for a software company can
nevertheless thrive.  It's not a business model, it's an operational
model. The worth to its users is greater than its worth to a single
proprietary company.

   When Mr. Mundie asks:

	2.Should an information-based economy protect the
        intellectual property assets that are driving its
        growth?

he's missing the point that the ``information-based economy'' for
which the answer is `yes' comprises only software companies.  When
``economy'' is understood to take in =all= businesses, the answer
frequently becomes `no'.  He actually alludes to this when he points
to ``the shift of focus away from the technology IP to content IP''.

   The only way a company can hope to continue making the big bucks
from ``technology IP'' is to =own= that IP.  So long as protocols can
be independently implemented, such a company is at risk of losing
customers to a clone.  (Watch for a push to outlaw reverse engineering
generally.  We already have an attempt to do that for encryption
methods, in the DMCA.)

	 [The GPL] also fundamentally undermines the independent
	 commercial software sector because it effectively makes
	 it impossible to distribute software on a basis where
	 recipients pay for the product rather than just the cost
	 of distribution.

Bingo!  He's got it, but can't accept it because it threatens his
business model exactly in proportion to how much it helps other
businesses.  GPLed software is worth the big bucks a maximum of once.



		 Best wishes,

			    Max Hyre

From:	 David Kastrup <David.Kastrup@neuroinformatik.ruhr-uni-bochum.de>
To:	 letters@lwn.net
Subject: Open Source and Forking
Date:	 Sat, 5 May 2001 02:36:50 +0200


Mundie from Microsoft has told us that Open Source carries the danger
of leading to forked software.

Open Source pundits tell us proudly that few examples of serious
forking exist, presumably because of the discipline of Open Source
programmers.

Both are way off the mark.  The question is: who wants to fork code in
the first place?  It turns out that individuals not out to make fast
money are not interested in forking third party code, or even working
with it.  Sad witness to this fact are, for example, literally dozens
of independent Web browser projects with different feature sets and in
different state of progress.

In almost all cases, the incitement to forking is only there for
commercial entities.  This is essentially what happened to the BSD
code base: the free base remained strong, and every company rolled
their own specialties.  Forks all around, and exactly because all of
these companies were able to protect their added value, their
intellectual property.  All but a few have died since, because the
cost of maintaining a separate fork beside a prospering free tree is
high.  This is the reason for proprietary Unices collapsing under the
impetus of the currently available free Unices.

So what does this tell us?  Forks rarely have a future in Open Source.
Even where proprietary forks are allowed (as with a BSD license),
natural selection tends to kill them off.  Where the incentive of
property is absent in the first place (such as with the GPL), forks
are even more rare.  Most of them have remerged at some time (such as
the gcc/egcs fork).  Only the strongest projects have a chance of
keeping more than one branch alive after a fork.  One of these rare
cases has been the Emacs/XEmacs split.

So it seems that Open Source does not lead to forking, and voluntary
programmers are not interested in forking either.  They either want to
help improve an existing project, or roll their own.  The only reason
for forking is to make money off your additional invested work by
keeping your branch proprietary.  So a license like the GPL is about
the strongest imaginable measure against forking, whereas a BSD-like
license relies on the power of natural selection to let only the
worthy projects survive and thrive.

In short, forking is about the least of our worries.  Total
duplication of effort is much more prevalent.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum, Germany
Email: David.Kastrup@neuroinformatik.ruhr-uni-bochum.de