Venting my frustration with FreeBSD

[Josh Paetzel]

10 years ago I thought it would be fun to upgrade my 486 from DOS 6.22 
to windows 95.  It took me about two weeks to realize what a horrible 
mistake I had made, and my general dissatisfaction led me to seek out 
a non-MS alternative.  I talked to some friends and mentioned that it 
would be sort of fun to learn a unix of some sort.  One of them 
recommended FreeBSD and about a week later I had a set of CDs from 
Walnut Creek in my hand containing FreeBSD 2.1.5.  I've been using 
FBSD ever since on my desktop and have even managed to make a living 
at administrating FBSD boxes for others.  I've run into other OS's 
over the years.  IRIX, Solaris, several linux distros, and am always 
glad when I can get away from them and back to FreeBSD. :)

This email was sort of prompted by the news that the friend who showed 
me the wonders of FreeBSD has started migrating everything he runs to 
linux.  I won't get into his reasoning, it's not really relevent to 
what I'm going on about.

I started thinking about what would force *me* into the corner of 
having to move to a different OS and came up with a short list.

1)  SMP scalability.  4-way boxes are relatively common, and hardware 
with higher CPU counts is only going to get more and more common.  
I'm no industry expert, but 5 years from now will my clients be 
considering buying 32 and 64 way boxes?  Possibly.  Will FreeBSD be 
in a positiion to compete favorably vs. the alternatives on such 
hardware?

2) RAID controller support.  This is a huge one that affects me 
directly even today.  Lack of in OS management tools for RAID 
controllers.  I have some options if I can pick the hardware, but if 
a client brings me something and says this is the hardware you have 
to deal with a lot of times putting FBSD on it means living without 
management tools for the RAID controller in the OS.  What good is 
hot-swappable drives if I have to take down the OS to rebuild the 
array?

3) Lack of direction in the project.

The FreeBSD project has lost it's focus.  10 years ago there was a 
clear direction and purpose to the OS.  It was targetted at 
high-performance network server apps on x86 and alpha hardware.  Now 
days I'm not so sure.  Is FBSD targetted at network servers, at 
desktops, at embedded devices?  What architectures do we target?  
Looking at the website I see alpha, amd64, ARM, i386, ia64, MIPS, 
pc98, ppc, sparc64, sun4v, and xbox.  I know this is a volunteer 
project, I know you really can't keep people from tinkering with what 
they want to tinker with, but xbox?  It seems to me to be a waste of 
resources to concentrate on anything besides i386/AMD64.....there's 
plent of market share to be captured right there.  I look at that 
list and see a future for AMD64.  (Yes, I wouldn't be at all 
surprised if sparc went away)  ARM isn't going away, but is FreeBSD 
really concerned with the embedded market?

FreeBSD does not have the support or the financial backing to be all 
things to all people.  It can't compete across the board with linux, 
so why try?  Even if FBSD got to the point where it was as flexible 
as linux, HI, I'm a hard real time OS, and I'm a terrific desktop OS, 
and I make a great platform for apache, and I scale really well on 
1024-way clusters, it would end up being linux.....fragmented and 
hacky and patched and ugly.

This is a rant, and I've purposely posted it here and not forwarded it 
anyone who matters because it's not going to make any difference 
anyways.  Feel free to prove me wrong, or to flame me.  It's slow at 
work and I could use the diversion.

If you *do* decide to flame me please take a moment to grep for 
josh at tcbug.org through the ports tree, or look for PR's with my name 
on them, or browse through the questions@ mailing list archives 
looking for responses from me.  I have, and do, contribute to 
FreeBSD, which I feel gives me the right to complain a bit.  I fully 
intend to ride the FBSD boat as long as possible, I just can't help 
but wonder if the slow leaks I see now are serious.

-- 
Thanks,

Josh Paetzel