Assuming We Want FreeBSD to Grow: Who Is It For?

[Anthony Atkielski]

Shawn Harrison writes:

> If the primary use of FreeBSD is for servers, then anyone who runs a
> server is our target.

Yes!

> I know most of the talk recently has been about big businesses and
> people who spend tens of thousands of dollars on hardware. Yes, it 
> includes them.

Yes.

> It also includes Joe Family Man or Jane Small Business who wants to set
> up a website. Sure it does. Why should Joe and Jane pay $35 a month (or
> even $10) for dippy Windows web hosting when they can use the DSL or 
> cable account they already have, and the "obsolete" computer from two 
> years ago, to run a top-notch web server using FreeBSD?

Exactly (provided that Joe Family Man or Jane Small Business are highly
computer-literate, that is). One can do almost exactly what the hosting
companies do with just a PC and a good broadband connection with a fixed
IP address, and it's generally cheaper than having a site hosted. The
biggest cost is telecommunications, not the server ... thanks to
FreeBSD.  The only advantages to hosting companies are greater total
bandwidth and more reliable and performant hardware, and these are not
usually very important for small sites.  I service 350,000 unique
visitors a month from the FreeBSD machine sitting on my desk.

Of course, Windows is prettier, and someone with absolutely no
experience running a server might prefer Windows, at least at first
sight.  But Windows costs $900 or more, and it requires far more
hardware to achieve the same performance goal, and the hardware must be
compatible with the OS, which is less likely with Windows Server than it
is with FreeBSD.  And even though one can get away with running a server
in ignorance under Windows, the fact is, anyone running his own server
is just asking for trouble unless he learns about the responsibilities
of running his own site, so Windows would only give him a sense of false
security.  To run FreeBSD he has to learn a lot more about running his
own Web or mail server ... but those are things he needs to know anyway
if he wants to do it all himself.

> Just think, they get a local network file and print server
> out of the deal.

More than that.  They get a Web site, a print server, a file server, an
e-mail server, their own DNS, their own NTP time sync, and potentially
more if they want it.

> I know it's probably too hard for them to do that currently, but that
> is just a detail that can be solved with writing documents and
> scripts.

I don't believe in simplifying things too much for people running their
own servers.  There are certain things they _must_ know if they are
going to take responsibility for their own server, and writing scripts
or other tools to allow them to proceed in ignorance is only leading
them down a dangerous path.

For FreeBSD, they must learn a lot more than they'd have to learn with
Windows, but the things they learn are things that any sysadmin owes it
to himself to know already, so forcing them to be aware of what running
a server implies is a good thing.

> It's easy to underestimate the numbers of such people. In every
> community, there is a significant percentage of technically "aware"
> people who would be interested in trying something like the above
> scenario, especially if they understood the results that they could
> get. I know a bunch of people who have tried setting up their own
> (Linux) boxen, with varying success or (more commonly) failure. How do
> we communicate with those people (1) what they can do and (2) how they
> can do it (3) and how it will be better in FreeBSD?

It requires a lot of heavy public relations, which in turn requires a
lot of time and/or money.  Linux is hyped by people who hope to make a
fast buck off Linux, and by people who are so driven by their hatred of
Microsoft that they are willing to spend every waking hour promoting
Linux.  I don't know if other operating systems without fanatics or
deep-pocketed promotors to support them can achieve that level of hype,
but it should be possible to do some PR at minimal cost.

Encouraging sites that run FreeBSD to actually mention the fact on their
Web sites would help a lot, especially if they link back to the FreeBSD
site (as I do).

On this point, it would be very nice to have a selection of official
FreeBSD logos in many different sizes that people could put on their
sites.  The complicated, multicolor graphics that are currently
available are too busy and difficult to read for many sites--they look
rather amateurish.

> If servers are primary, perhaps a secondary use of FreeBSD is the
> desktop. Well, yes it is. We have X in the ports tree, and a couple of
> different canned installation options right in sysinstall that will give
> you an X desktop right off the bat. So it's part of the system, whether
> you agree that it should be emphasized or not. Perhaps Joe Family Man or
> Jane Small Business, or even Mr. I. M. InfoTech Manager ("That's IM^2 to
> you!") will install a FreeBSD server for some purpose, only to discover
> the desktop and try it out.

There's no problem with that.  If they can run a server correctly, they
know a lot about computers, and if they know a lot about computers, they
should know what they are getting into if they decide to run UNIX on the
desktop.

> I have my computer-illiterate wife and nine-year-old daughter using KDE
> on FreeBSD, and they're both very happy with it -- much more than with
> Windows 98. My wife can finally build her website simply by saving 
> files, and my daughter can play Mr. Potato-head and draw pictures. So 
> they're both happy. And I'm happy having that same box serving web 
> pages, a database, and mail for my family.

All well and good ... but they didn't install or configure the desktop,
did they?  Having a UNIX desktop in the family requires at least one
computer geek to install and maintain it.

Certainly if you have the geek available, and if a FreeBSD desktop does
what you want, you can install one, and save money while improving
security (because UNIX desktops are rarely targeted by viruses, outside
the category of Linux and OS X).

The only problem I have is with promoting FreeBSD (or any version of
UNIX) as a drop-in replacement for Windows.  It will never be that, and
it is very misleading to suggest to anyone that it might be.

FreeBSD is not a drop-in replacement for a Windows server, either, but
the situation is different for servers, because someone running a server
cannot afford to be computer-illiterate.  A server admin has to know
what he is doing, and if he knows what he is doing, he can run FreeBSD
just as easily as Windows.  Indeed, if he knows what he is doing, he
will likely spontaneously decide that FreeBSD is preferable, anyway,
since it has many advantages in a server role.  That's what happened to
me ... after trying both FreeBSD and Windows on the server side, FreeBSD
was clearly the winner very early in the game.

> There is no inherent conflict between growth in the desktop and the
> quality of the server codebase.

There are conflicting requirements for servers and clients in the code
base.  You cannot handle both with good results.  You can do both in a
very mediocre way, or you can do one very well and the other poorly.
Since FreeBSD will never seriously compete with Windows on the desktop,
it's irrational to try to emphasize desktop use, because the desktop is
a battle that cannot be won, and any emphasis on the desktop is to the
detriment of the server side ... a battle that FreeBSD is _already_ in a
position to win.

> If FreeBSD doesn't welcome people who are interested in working on
> desktop issues or device drivers for consumer scanners, those people
> aren't going to take their efforts to working on the kernel. They
> really want a device driver for that scanner, or a GUI console for
> printing. They'll take their efforts to Debian* where they'll get some
> recognition and support.

Good.  They can have it.  I need a reliable server, not a kinda-sorta
server that wants to be a desktop.

I don't see Sun trying to position Solaris as a Windows replacement on
the desktop.  I don't see IBM trying to position AS/400 on the desktop.
Ever wonder why?

The only vendor that can challenge Windows to any extent at all on the
desktop is Apple. Yes, their latest OS has UNIX under the hood, but
they've put billions into a GUI that hides it all, and they've done a
vastly better job of this than the ratty little GUIs that are used on
other open-source UNIX systems. Not only that, but their GUI is a GUI of
its own ... not a wannabe Windows. The difference in quality on the
desktop is orders of magnitude between, say, Linux and OS X.

You might also say that people like Sun have nice desktops.  Yes, they
do, but they don't try to claim that a Solaris desktop is a realistic
substitute for a Windows desktop.  A handful of users can get by with a
Solaris desktop, but most need Windows, and Sun is never going to get
rich by promoting their OS on the desktop or by trying to make their OS
behave more like a Windows desktop (especially if it impacts the use of
the OS as a server).

-- 
Anthony