Become a fan of Slashdot on Facebook


Forgot your password?
BSD Operating Systems

FreeBSD 4.1.1 vs. Linux 2.4 623

A reader writes: "This article finds's Linux guru wondering why he isn't running FreeBSD. 'Linux 2.4.0 is available for no money. So is FreeBSD. Linux uses advanced hardware, so does FreeBSD. FreeBSD is more stable and faster than Linux, in my opinion. We penguinistas sometimes believe we are having more fun than anybody. But then I lean over the fence and discover the FreeBSD folks are having a hell of a party, too. And their OS is as fast as I have seen. I have to ask myself why I don't just switch my server to FreeBSD.'"
This discussion has been archived. No new comments can be posted.

FreeBSD 4.1.1 vs. Linux 2.4

Comments Filter:
  • All the BSD documentation you need comes with the installation. I'm not kidding. The FreeBSD man pages are complete, comprehensive and the number one reason why I run it over Linux.

    Linux documentation s u c k s.

    Now, if man pages arent enough for you then, sure, Linux-My-First-Day-With-Linux-HOWTOs are for you.
  • Actually, pkg_version -c will tell you exactly what needs to be done to upgrade a port.

    Here's an example from my system...
    pathwalker% pkg_version -c

    # slrn
    # needs updating (index has
    cd /usr/ports/news/slrn
    make && pkg_delete -f slrn-
    make install

  • There are really only two reasons I use linux, and they're not particularly technical. In technical terms I consider them to both be powerful modern unix's with rapidly evolving perfomance and facilities. BSD has a bit more heritage, Linux may have more developers, but I could probably be happy on either.

    My first reason is the silly one. In the early days of Linux the *BSD crowd (plus SCO) were amazingly smug and arrogant. Very much, "Why are you weenies playing with your toy? No wonder you can't handle BSD.". Sure, this wasn't the whole community, but the BSD camp (at least here) was
    full of "serious sys-admins" with this attitude.
    Perhaps Linux has actually helped BSD realize that attracting new users is a useful part of an OS community?

    The second part is focus and energy. And to an extent it's related to the user base. BSD people (used to?) focus heavily on servers and tweaked code. Interested in optimisation of performance and stability rather than agressive experimentation. I love the linux communities willingness to make agressive core changes and happily tilt at windmills. When Linus said, "let's target the desktop" most people thought he was totally nuts. But now with things like Gnome, KDE unix actually exists again as a reasonable, and improving, desktop. Likewise linux pushing to PDA's, super-computers,'s all very exciting.

    Sure, BSD is capable of all these things as well. But I still think it was Linux that set the pace and forced BSD to come out of its shell and match it.
  • Come on already.

    If you haven't figured it out, I'll let you in on a secret: the various BSD distributions excel at non-desktop uses. Unless you really *ARE* an 31337 h4x0r, there is not much point in running them on the desktop. Right now I'm using Windows on my desktop, in fact (although this is primarily because my Thinkpad's screen is hosed -- I do prefer Linux).

    Meanwhile, if you run Linux in a very high-load, high-availability environment, I can almost guarantee that you will have more little problems than with FreeBSD. When every little problem turns into a page or an email in the middle of the night, you have to become a little wary of
    Linux. I like Linux. I hack on Linux. I sure as hell don't run web server farms or firewalls on Linux. Choose the right tool for the job.

    Your argument about books in the local bookstore is also spurious. It indicates that your bookstore isn't very good. Look for 'The Design and Implementation of the 4.4 BSD Operating System' or 'Building Linux and OpenBSD Firewalls'. All of these are well written and illustrate what's going on. I'll go so far as to say they do it better than any of the Linux books do. Look in the canonical Unix sysadmin's handbook, the purple 'Unix System Administrator's Handbook, 3rd edition'. What representative OSes have they chosen? FreeBSD, Linux, HP-UX, and Solaris. That's pretty mainstream, dude.

    I don't think you're wrong to use Linux instead of *BSD on the desktop. I think you're wrong to compare them at all. BSD is for servers.
  • And I've been saying it for years.

    In every single instance where I've noted a difference between the BSD utility and the corresponding GNU utility, I've preferred teh BSD way. The options (none of that --fifty-charactar-option-string-instead-of-two, the information (e.g., ftp status line), man pages instead of that moronic info system, etc.

    A linux kerenel with BSD utilities might interest me (but why bother?), but certainly not the other way around.

  • That certainly explains the relative popularity of Linux versus the *BSDs.

    People do infact care if their sweat turns into corporate welfare. They might not care to protect their code in the most agressive manner possible. They don't need to.

    That's something that gets glossed over by the pro-BSD camp. Most Free Software coders are more moderate than RMS and are accomodated well by the available licences.

    All the BSD licences do is make it easier for Robber Barons to extract the equivalent of corporate welfare out of the generosity of Free Software coders. OTOH, the LGPL allows component sharing and the perpetuation of standards without giving corporations a free ride when it comes to embrace and extend (Winsock anyone?).

    Most reasons to favor the BSDL over the LGPL are highly suspect. Also, there is usually no real need.
  • You cannot license reference implementation under GPL, as far as I understand, since GPL is not a free license.
  • The Debian Linux distro is supposed to become kernel agnostic, at least between the Linux and Hurd kernels. You'll be able to use either kernel in the relative comfort of your familiar Debian environment. Would it be so hard for someone to take that in the *BSD direction as well?
  • a xxx 666 (the xxx were actual letters) license plate, maybe I should switch to BSD

    pictures, please...
  • The same can be said of FreeBSD and Linux with regards to desktop applications. Stick with your strong points.

    I almost agree with you - I don't see the need for (most) servers to have a sound card. However, I disagree with your idea of using FreeBSD only for a server OS.

    Recently, my wife's hard drive took a nosedive. It was no big deal, because her /home was NFS-mounted from our FreeBSD LAN server. All she lost was the Debian installation she'd been using, and a quick (well, as quick as a network install over ISDN gets) re-install and she was back up and running.

    I decided to try an experiment, though. I wiped the freshly-installed Debian and replaced it with FreeBSD, adding the appropriate packages for X, gdm, Netscape (the Linux version running in compatibility mode), WindowMaker, etc. Know what? She didn't notice a thing. Her desktop looked, acted, and sounded exactly like it did before.

    The only think she eventually discovered was that her mouse no longer gets jumpy during heavy loading - all of her apps even stay responsive (if slower) until the load goes back down.

    I asked her if she wanted Linux back. She said that she missed having a little stuffed Tux on top of her monitor, so I bought her a stuffed Chuck. Now she's completely happy with her rock-solid desktop which never goes down, bogs, or otherwise interrupts her usage.

    I like FreeBSD for the desktop. Almost every app you'd ever want is available from the ports collection, and almost every Linux app runs perfectly (except for the ones requiring kernel modules that haven't been ported yet). What more could you ask for?

  • Debian isn't particularly hard to get up and running, and once it's up it is so easy to administer. Did someone find a security hole in BIND this week? Fine; with a single command you can get and install the fixed version. In fact, with a single command you can get all the latest stuff. (The command is "apt-get upgrade" and you can specify which mirror it should get the stuff from; I use the University of New Mexico, for example. You can choose from "stable", "testing", and "unstable"; for servers you would probably only use "stable".)

    That seems pretty nice. It is a little more complicated in FreeBSD.

    1. Run cvsup to get updated sources and things. The config file you need is described in the handbook "cvsup config-file" and clicking on "get" will do, it can be run without X as well.
    2. cd /usr/src; make world
    3. cd sys/i386/conf
    5. cd ../compile (or something like that; conf will tell you the exact location)
    6. make depnd&&make&&make install
    7. mergemaster
    8. reboot

    The middle bunch of steps on making a new kernel are optional, only needed in the unlikely event that you want something not compiled into the GENERIC kernel, but recomended as you seldom need anywhere close to all the stuff that is in GENERIC.

    By following the handbook directions you can get the stable branch, the bleeding-edge branch, or one of the numbered releases, or a stable branch for a older major version. There are a few other choices as well.

    They still have a chance at winning me over: they just need to code up a BSD version of apt-get. (This implies Debian-style packages... does BSD even have packages?)

    It has packages, and ports. Mostly just the ports are enough. They are trivial to use (for example "cd /usr/ports/lang/smalltalk; make install" will fetch source for smalltalk, and all the packages it needs, patch them and install them).

  • I'm afraid I'm another one of those apt-thing bigots.

    If all you are afaid of is how to replace apt-get and apt-install, that one is easy. Try "cd /usr/ports/[catagory]/[thing]; make install". That's it. It'll find any dependent packages, and make them, download source from any of N place, apply and FreeBSD patches, and do the install.

    /usr/ports is one of the coolest things about FreeBSD. I do have to admit that mergemaster isn't too bad either.


  • Some kind of a meta port that installs a userfriendly desktop along with a standard set of apps would be super cool.

    Have you looked at the /usr/ports/x11/gnome "meta-port" for the GNOME integrated X11 desktop? Seems like what you want. I assume there is a KDE one as well.

  • I would love to go on, but if you have never really spent time using/upgrading/maintaining Debian systems, you just have no idea. There is no point in trying explain it.

    That's a copout. I've been thinking of trying a Linux. Convince me Debian is the one.

  • Great, but what about updating all the ports software you installed?

    I don't know of a way, there might be one. I havn't done many source upgrades, so it hasn't been an issue until lately. I havn't checked the handbook.

    What about preserving configuration files?

    That is that the mergemaster step does. Deals with conf files you havn't touched, and gives you diff's for the rest.

    What if you don't want to "make world" a whole distribution? Compile? Thats a joke.

    I like to compile my world. I like to know for a fact that if I patch someting and recompile the only change is the patch.

    What is the point of having a distribution if you have to compile all your updates/additions?

    It is faster for me to do that then download a new install image, and I have a 256K Frame-Relay, and a relitavly slow CPU. It seems like a better tradeoff to deal in compressed source then whole binary images.

    It would seem that the FreeBSD way is more that a little more complicated. Many of the features just are not there. But thats Ok. Its probably one of those things where you don't need it until you have it. I mean, I was perfectly happy with COMMAND.COM, freeing up conventional memory, allocating XMS memory, and programming using SEG:OFFSET addressing under DOS before I first installed Linux. Sure, DOS can use all 64MB of RAM in your computer, its just a little more complicated :)

    Well, feel free to tell me how the other ones does it better. I wasn't saying FreeBSD is the ultimate, just that it ain't bad. I know for a fact it's VM system isn't as good as NetBSD's (NetBSD's UVM rocks, even if it needs to be tuned), it isn't as secure as OpenBSD, and you can't buy support for it as easily as BSD/OS. I've even heard Linux has a thing or two it's good at too :-)

  • Maybe one is faster than the other in some things, but so what?

    And, besides, often "X is faster than Y" is better phrased as "release N of X is faster than release M of Y" - release M+1 of Y may well catch up to, or beat, release N of X (and then release N+1 of X may surpass release M+1 of Y, and then release M+1 of Y may surpass release N+1 of X, and so on, and so forth).

  • First of all I think this benchmark is unfair. Even that being the case linux showed it can outperform Freebsd in mysql and sendmail. I believe this test was unfair in the following ways: 1) The author used the latest version of Freebsd, yet did not specifiy what distro of Linux he used. For this to be really fair, I think he should have used say RedHat 7.0 (with patches) (or any other distro based on glibc 2.2 compiled with -i686 extentions). 2) There are a number of well documented performance (bad paging) and stability issues with Kernel 2.4.0 which all have been corrected in 2.4.1-AC3. Because of this I think it would be more fair to either test 2.4.1-AC3 or wait until 2.4.2 and retest. 3) He should of used apache 1.3.17. 1.3.17 finally adds some Linux specific hooks to take advantage of v2.2+ kernels (see for details: Now before the Freebsd crowd jumps up and down, I ask you to check the apache history file for all the Freebsd kernel hooks in 1.3.x apache compared to linux. On top of this, it would seem that the Linux kernel 2.4 adds even more functionality that apache could take advantage, but at this time does not. 4) I would like to see if using the version of Mysql and sendmail that comes with redhat 7 changes the results at all. These applications have been compiled with the 2.96 compiler which is supposed to effect performence to some degree. In summary I don't think this was a apples to apples comparision as the author used the latest and greatest Freebsd (not just the kernel) against an unknown Linux distro. I would be interested to see if Linux couldn't also pull ahead of freebsd in webserving with Kernel: 2.4.1-AC3, apache 1.3.17 and glibc2-i686.
  • It's very common among people who have been used to being 'superior' to their peers for as long as they can remember. Great athletes, unusually intelligent people, great artists etc very often turn out to be arrogant assholes, with little or no patience for people they consider less talented. It is just a bit more pronounced in good coders, because a highly analytical mind often comes with a lack of interest/talent for empathy and social interaction.
  • by rve ( 4436 )
    The csh has been replaced by tcsh in the stable releases, but on all my boxes I replaced I use a statically linked version of bash as rootshell.

    Drivers DO cause problems sometimes. The PCM driver for instance worked perfectly for mostowners of a SB-Live card, just on my particular hardware configuration it caused a kernel panic. Recently fixed in release 4.2. This kind of bugs are spotted and thus fixed sooner with moer widely used OS-es like linux.

  • let me ask you this, how do you plan on getting the source to sed on a debian,redhat,slack,etc system? Much easier on FreeBSD system.
    Umm... apt-get source sed? The level of ignorance in both these camps is ridiculous...
    P.S. Honest question: How good is /stand/sysinstall? Can it do specific packages? Are all packages available via /ports available with /stand/sysinstall?
  • As for socialized medicine, I would say that the reason is stupid mothers getting high on crack or having abortions rather than the health-care system. Really. At my job I make $30K, when my son was born, he had some problems, and we got to _choose_ which hospital to take him too. Not which hospital in town, but which hospital in the whole U.S. That doesn't happen with socialized medicine.
  • I have never seen anyone say that they use BSD because of userland stuff, except possibly for the initscripts. Even if it is part of what BSD _is_, doesn't mean its part of what makes BSD good. I actually think that the userland part of BSD is what is holding back its acceptance somewhat. Really, when someone installs a sun box, what is the first thing they do? Install the GNU system. Why? because the tools that ship aren't nearly as user-friendly. The GNU userland is the best UNIX toolset I've ever worked with.
  • I'm guessing that's because you don't hammer professionally. If you know a good carpenter, you'll find that they have tons of similar tools, each with a special purpose.
  • 1) Linux is actually starting to emerge on most of these platforms

    2) Linux can generally take better advantage of the hardware on the machines it runs on. NetBSD may support more architectures, but Linux supports more architectures well. That's probably just because of the number of people running it.
  • If you are running a server, when does it have time to be compiling packages?
  • You're missing the point. With a binary distribution mechanism, you can simply run the update, which takes 10 seconds, and then restart, which takes about 2. So, you are basically degraded for 10 seconds, and down for two. With source-only distribution, you've got to spend lots of time compiling. Depending on the package, this could take hours, especially on a loaded server. So, instead of measure degraded performance in seconds, its hours.
  • If you're having difficulty with the installer, I suggest taking a look and ensuring that all your hardware is on the supported hardware list ( []). FreeBSD can definitely be picky about the hardware that it runs on. Also, you might try doing just a basic install, then cvsup'ing to get everything else.
  • I've often thought that FreeBSD could be a super easy Desktop OS with little work.

    Everytime someone mentions FreeBSD on /. people have to mention the ports collection and how cool it is. Then, all the Debian people counter with how cool apt-get is. Both are very cool, but I have to say that FreeBSD continually impresses me with how complete the ports are.

    Some kind of a meta port that installs a userfriendly desktop along with a standard set of apps would be super cool.

    I've always had an easier time of getting hardware working in FreeBSD than Linux, including soundcards and my video capture card (just add "device pcm" to the kernel config file for 99% of sound cards)

    In my years of running both FreeBSD and Linux, I've only had one hardware snafu with BSD (if anyone can help, it's the integrated sound on the FIC AZ11 Mainboard. It only plays noise.:-( No IRQ conflicts that I can find)

    I'd love to package a FreeBSD/GNUstep [] port with a nice installer for FreeBSD, but I lack a lot of programming experience. If anyone knows of a project similar to this, I'd love to hear about it, or if not, someone might be interested in forming a group?

  • Godwin's law ain't named Hitler's law, so let's not inadvertantly honor Stalin in any way. How 'bout Maxwell's law?
  • "Yeah and i'm sure Windows will continue to be improve too."

    The problem with Windows is that it continues to get better and get worse at the same time.

  • Well, the 4GB cap on memory with an Intel processor on Linux is causing me some headaches at the moment (massive database stuff). If FreeBSD can do 768GB, I'm so there.


  • Being an open-source developer myself, I would personally never release any of my code under anything but the GPL. And I encourage all the coders I talk with to do the same.

    I see your point about making your code as free as possible, but I also don't want to see someone else taking my work, "embracing and extending" upon it, and then profiting off a piece of software that I originally intended to be free for everyone. The GPL guarantees that what I intended to be free will stay free.

    And yes, communism has thankfully failed with politics, but socialism is doing quite well. Check out most of Europe, and then ponder why America, as the only developed nation without socialized medicine, has a higher infant mortality rate then Thailand.
  • You use pkg_delete? I have always just installed one port over the old. Why? Because once you cvsup the port, make deinstall fails to work. pkg_delete and all the pkg_* commands are a pain because they demand the exact *version* of the package in the name. Basically I use ports as an autobuild mechanism, but the package management may as well not exist. Funny thing, my system hasn't keeled over for it yet. I don't miss having to --nodeps every other RPM on my redhat box or deal with debian's weird notions of meta-packages like that perl debacle (which is still about as confusing as can be).

    I usually recommend freebsd to people who like slackware. I call it "slackware done right". You get all the benefits of manual control and compiling everything from scratch, with lots of automation for the rote tasks -- like compiling everything from scratch.
  • Dated? The ports collection almost invariably tracks the very same sources Linux compiles from (and often has to apply patches to remove linuxisms from the code). Granted, ports aren't updated as quickly as linux packages are made, but it's a stretch and a half to call them outdated. You *have* updated your ports tree with cvsup, haven't you?

    And isn't it nice BSD will run just about anything Linux can? Quake III for Linux runs just as fast under BSD as Linux (some say faster).

  • Debian has tossed around the idea of a BSD distribution, but the idea usually dies off quickly. Debian GNU/BSD is unlikely. Although the BSD license gives Debian every right to fork the codebase and GPL it, it not only be a public relations disaster for Debian in garnering the ill will of the BSD developer community, the split would become technically enforced as well, since the GNU fork would find itself having to backport any changes made to the BSD codebase. I don't think Debian would be stupid enough to GPL a distro of BSD.

    Yes, BSD has packages. Not great, but serviceable. BSD doesn't need apt, it has ports, which is based on a perfectly good dependency-management tool with support for pre and post-installation and hooks at any point of the build and install process. It's called make.
  • > I installed freebsd on a box and I couldnt get the newest version of apache for it...

    /usr/ports/www/apache13 contains apache 1.3.14. I haven't run cvsup since 17 was released a week ago (which immediately followed 14, there was no 15 or 16). Checking, i see that the current apache package (and therefore port) is indeed 1.3.17. apache is compiled from the exact same codebase as apache for linux. it is likely to be bug-for-bug compatible.

    /usr/ports/www/mod_php4 contains PHP 4.04pl1, exactly the current version of PHP.

    sshd is configured by default to disallow direct logins of root, you are expected to su. This is fixable, one starts by reading the documentation for sshd. Don't mean to be snippy, but that's a feature.

    ISO images of FreeBSD are available at O-IMAGES/. Personally I install from FTP, but I've always had mad bandwidth whenever I've installed.

  • Just a note: step 5 and 6 have changed, they are now:

    5. cd /usr/src
    6. make kernel KERNEL=YOUR_KERN_CONF
    7. make installkernel KERNEL=YOUR_KERN_CONF

    One neat thing about FreeBSD's process is that it makes it easier to keep multiple kernel configs around. Linux's lets you save a kernel configuration file, but it clobbers an old build of a different configuration when you use the new configuration.

    oh btw, you kinda forgot make world :)

    apt-get upgrade is still pretty neat though, cvsup is nice for syncing the sources, but not too useful at telling you what's changed and upgrading single ports or pieces of userland at a time.
  • > Some kind of a meta port that installs a userfriendly desktop along with a standard set of apps would be super cool.

    You mean like /usr/ports/x11/kde2 and /usr/ports/x11/gnome? They're there. Takes freakin forever to compile it all, but they're there.

    GNUstep already exists for FreeBSD BTW. /usr/ports/devel/gnustep. Not too useful by itself, better to just install the windowmaker port and it'll install gnustep too.
  • If they didn't GPL the kernel, just perhaps the term GNU /BSD is a little mistaken. God damn people, read what I say for once.
  • This is precisely why I won't use any of the BSD variants. Their persistent belief that somehow, their licence is a good idea and that the GPL is a bad idea.

    Saying the BSD licence is freer than the GPL is like saying a place without laws against kidnapping is freer than a place without.

  • Saying the BSD licence is more free than the GPL is like saying that a state without laws against kidnapping is more free than a state that does have them.

  • People have been taking code under the BSD licence, doing essentially nothing to it, spend tons of money on marketing, and reap the rewards. The only thing that brought free software out of the closet that those who would steal from the public domain shoved it into was the GPL. If it weren't for that licence, nobody would even know free software existed.

    Your argument is ridiculous and stupid.

  • 2 points:

    a) You can always get the newest version of apache and php for bsd. Get a current version of the ports tree(use cvsup), cd /usr/ports/www/apache;make install. That's it. The current version is available within several days of the release.

    b) You can't ssh in as root because it's a *really* stupid thing to allow people to do. The BSDs, IMO, come with a far more sensible security setup than Linuxen and other SysV style Unices, e.g. group wheel, jail(), kernel security levels, etc. Security tends to make doing stupid things harder, that's pretty much the idea.

  • They all have their strong points, and their weak points. MS's various offerings are easier for people with little experience to use/admin. You can buy excellent support for Solaris. OpenBSD is very secure. NetBSD will run on just about anything.

    Both FreeBSD and Linux are attempting to fill the niche of "fast feature-full Unix". FreeBSD (at least by these benchmarks) is a bit better at the fast part, but I think that Linux is more feature-full, and has better commercial support.

    It really doesn't matter, anyway. Having two free Unixes that run well on PC hardware is better than just one. If every box out there was a Linux box, when (when, not if) somebody finds a new remote exploit, then everybody would be vulnerable. Similarly, an all-FreeBSD world isn't good either. A mix of different operating systems, all slightly better at their own little tasks, is more robust than a homogeneous environment. It was poor performance compared to NT that spurred many of the improvements in 2.4. It will be a wonderfull thing if 2.6 becomes better because FreeBSD's VM kicked the 2.4 VM's ass. It will also be a wonderful thing if FreeBSD can benifit from friendly competition with Linux (those mail scores seem a bit low. tsk. tsk. tsk.).

    Competition and cooperation between the free OS's will lead to better free OS's. Encourage both as much as you can. Just remember that FUD spreading or any other pissing contest is not healthy competition.

  • I'd be happy to give FreeBSD a try, but I'm only going to do it when it's distributed by Debian. Yes folks, I'm afraid I'm another one of those apt-thing bigots. I've no particular desire to go clambering up the FreeBSD learning curve for its own sake, any more than I'm interested in doing the same with other Linux distros. Debian FreeBSD sounds like a fine idea to me, though.
  • It's insurance that control of the free OS world won't be captured by some corporation(s).

    Please explain how a corporation can 'capture' *BSD? Does the code suddenly disappear when a corporation distribute a binary without releasing the source?

  • Couple of points (which seems to fly over your head):
    1) One of the points with free software and napster is that me using/sharing the music/software has no effect on you. You don't lose anything by me sharing the music because you can still sell your copy. I keep hearing this argument cropping up again and again. Does it dawn on you yet what I'm trying to get at?
    Microsoft using the BSD TCP/IP stack has no effect on *BSD. The code doesn't magically disappear because Microsoft use it.

    How is this different from Redhat using GPL code? Or is the objection mostly because Microsoft is using the code?

    In both cases, you have commercial entities using code, in such a way which the license allows them too.

    instead of coding themselves, it took less time to take someone else's free code, change the license and sell it.
    2) On the BIND thread, some people are suggesting forking the code and slapping the GPL on it. How is that different? Both times you go against the implicit wishes of the original author.

    If it was Linux code it would be called theft
    RMS calls it misuse, not theft.

    Do you even code, or are you mostly one of those cheerleaders who don't contribute anything but flames?

  • Soes Sun Microsystems ring a bell?

    Yes. Does UserFriendly ring a bell?

  • Yes, see Microsoft Windows TCP/IP.
    Where's the source?

    I'm confused. So, the BSD people had rewrite the TCP/IP when Microsoft started using the BSD TCP/IP? Or did the fact that Microsoft starting using the BSD TCP/IP have NO EFFECT on *BSD at all?

  • out of curiousity, what were the bottlenecks in the linux kernel when you were poking with the performance and scalability tests?

    i have found smb on solaris/sparc to outperform smb on linux/i686 with approximately equal horsepower (333 ultrasparc2 vs. pentium 2 450).

    i think it was the case that the memory bus and i/o bus were less latent AND had higher bandwidth on the sun than on the intel, but i never really got down to profiling it... (not to mention level 2 cache latency differences) all this aside, i had a feeling that it was actually none of these things (but rather the very efficient network stack in solaris 8) to blame...

    i'm very curious to see your assessment :)

  • "Yes, this is not a real benchmark. But at least it shows some evidence."

    I.e. Yes, it is not valid to draw conclusions from, but I'm going to do it anyway?

    Hint: Fake Benchmarks = spurious results (spurious means doubtful or unreliable).

    Trusting them is bad.

    Allow me to demontrate:

    I will use the following non-scientific perl code to benchmark:

    my $i;
    if(`uname -r` =~ /bsd/i) {
    for($j = 0; $j 100;
    print "The result is $i!\n";
    if(`uname -r` =~ /linux/i) {
    print "The result is some big number!\n";

    Then I time that under linux and *bsd. Strangely, it goes much, much faster under linux. Sure, I've never claimed that it's a good benchmark, but at least it does show some evidence that linux is better than *bsd.

    You'd think that people would be familiar enough with science that bad methods completely invalidate your data. (It stems from the fact that to be ignorant is better than to be wrong.)
  • vmware crashes more on FreeBSD than Linux assuming you get it to run..

    more games area available for linux that wont run to well under FreeBSD ...

    You have been caught in the hype of Linux, and missed the BSD boat.

    Mainly I use Linux at this point because of the package management. Sure you can say BSD has the ports, but that is not really package management like rpm or deb. I know that not everyone like package manangement, but it is actually a benifit to some of us.

    I do wonder why Linux does some of the things that it does. In particular why it is so common in the linux community to make it so difficult on upgrade. libc5 to glibc rpm 3 to rpm 4, etc. I hav ealso noticed that FreeBSD uses a better system configuration. Baasically on FreeBSD I am told that you have one config file that sets up weather or not to run X and weather or not to run inetd and all the services etc, where as Linux distros usually have so many in the rc scripts. This could be changed, but for some reason we like sys V style??

    I don't want a lot, I just want it all!
    Flame away, I have a hose!

  • Freebsd already has an apt-thing type thing. It's called "the ports tree".. for example:

    cd /usr/ports/graphics/Mesa3
    make install

    The Makefile will fetch everything needed to compile mesa, then build and install it. When you want to get rid of it, just make deinstall. And it's got a mechanism for updating the ports too. See the freebsd handbook for that though.

  • Interesting. I've often found the opposite. Whenever I have a problem I need to solve, I go to one of two places:

    The FreeBSD Diary [] or the FreeBSD Handbook [].

    I like the diary because it's a collection of problems/solutions that actual users have run into. It's very practical. The Hanbook is basically the official documentation for FreeBSD. I like the Handbook, because whatever it says, is the way it is. I don't have to worry about "Will this How-To work for my distro?", because there is only one FreeBSD.

    I've always had a bitch of a time finding the solutions I need under linux, probally because there are 90 million different ways to do things. That could be good or bad, depending on your viewpoint.

  • While I agree with your discussion about the BSD license vs. the GPL, the following statement is a bit snooty and highly inaccurate:
    Windows haters use Linux, but Unix lovers use FreeBSD.
    You are implying an elitism around FreeBSD that is just self-aggrandizing. I've loved Unix since 1987 at MIT, and I started using Linux in 1995 because it was a lot more accessible: it was far easier to install, get support for it, and in general seemed an easier beast to deal with. Yes, it is entirely possible to be a competent programmer whose development platform has almost always been Unix, and to prefer Linux over FreeBSD.

    Your statement implies that only geeks that care first and foremost for the technical superiority of every aspect of the kernel, and its most direct subsystems, can be "True Unix Lovers". That's just arrogance. Plenty of extremely talented programmers, who "get" Unix and love it over anything else, would prefer Linux over FreeBSD simply because accessibility, support, ease of installation, and software availability are far more important criteria.

    If the shoe fits, wear it.

  • I suppose you can argue this, but what is an example of software that doesn't require frequent updates and maintenance?
  • But the context of the whole discussion is software that is networked, so therefore security is an issue. It also implies that the software itself is somehow related to networking.

    perl -e 'print "Hello world\n"' does everything I need, and requires no frequent updates or maintenance, but it's not relevant to the issue of network security.

    My basic point is that any sufficiently useful, networked program is going to require updates and long term maintenance. Only time increases stability and security, but only asymptotically. At all times does the sysadmin need to keep a careful watch. There are no software exceptions to this; not Linux, not Windows NT/2000, not OpenBSD, not Solaris, not anything.

  • Switching solely due to the Ramen worm is a pretty sad and scary reason to switch. You might as well switch OS's at random every couple of months.

    The problem with the Ramen worm was not with Red Hat, but with sysadmins that don't frequently update and maintain their systems. If you don't do that, it doesn't really matter what OS you use.

  • FreeBSD may be better than Linux for what it tries to be (which is a lot).

    After a failed install of FreeBSD 3.1, I now have FreeBSD 4.2 running, though with a slight bit of trouble. However, 3.1 was a disaster, not because of the software problems that prevented it from being installed, but because of the attitude problems I encountered asking questions about it to resolve the problems. It is because of that I cannot choose to go with FreeBSD for project where it does seem to be technically superior.

    As in the commercial business software sector, the free open software sector needs support, too. And FreeBSD is more lacking in this area than Linux is (though both certainly are), at least from the perspective of someone like myself who is already more familiar with Linux. Support for free open software comes online in places like IRC (analogous to an 800 code toll free number for fast responses to simple questions), USENET, and various mailing lists. However, I found all of these had lots of people with attitude. As expected, IRC was the worst.

    Case in point. FreeBSD 3.1 didn't install on my system and stopped at an error message saying it could not find the CDROM (but it booted from it, and Linux, OpenBSD and Windows used that CDROM just fine). Lots of people said my hardware was broken (it was not). Lots more people said FreeBSD didn't support ATAPI CDROM on the IDE secondary master and followed that up with reasons why not, such as it was in violation of the standards (I went and looked, and no it is not). Ultimately it turns out the FreeBSD kernel works fine with such a CDROM; it was the installer program that failed. By the time I did find out, I was so disgusted by the people, I decided to not finish doing the install.

    I came back with the release of 4.2. I ran into a simple problem and I immediately knew a way around the problem. But I wanted to avoid that workaround if I could. The problem was I had planned for 8 partitions, but disklabel only let me do 7 (partition "c" not counted). I asked on IRC if this was indeed a real limit or if there was a way to work around it. I had lots of wrong replies, including one person who said that was a limitation of SCSI (odd, since I'm using IDE). The most common was that my hardware was broken. Lots of people replied "I installed fine with one partition" even though I never asked "How many partitions did you install with?". Eventually, about an hour after I first asked, and 3 channel visits later, one person (thank you Metrol) finally provided me with documentation that showed there was indeed such a limit. So I went and changed my strategy to accomodate the limit (including a re-install of OpenBSD the same way since I was comparing these). But it left me with a renewed disgust for the FreeBSD community.

  • Thanks for the invite. Next time I'll stop by there first. Hell I might even do so without a question just to say hi.

    I hope that when I ask a question, and someone knows that the answer does happen to be in a certain document, and realizes that these things can be missed, they would simply say where to find the answer.

    There are lots of things about BSD I don't remember, or never learned, when I used it many years ago (because I wasn't really doing all that much with it at the time). When reading documentation there are two ways to go about it. One is to read it sequentially like a book. This takes longer the bigger the documentation is (The FreeBSD Handbook looks like about 6 to 12 hours worth) and of course lots of gritty details get forgotten. Another is to try to match your own interpretation or wording of a concept with the index or table of contents. Sometimes that works and sometimes it fails. And then, even when you get to the right part of the document, the details simply go unnoticed because the wording is too BSD-ish (not unexpected in a BSD document, right).

    I'm very much a help-myself kind of person. Rarely would I ever need to ask someone for help. But what is useful is good, well organized, technical info. Usually the answer I look for is very succinct. Too often what I get isn't (even in the Linux community).

  • I was installing both FreeBSD and OpenBSD with exactly the same filesystem organization (e.g. which mount points are separate filesystems) and partition sizes as I normally use in Linux, for comparative purposes. I installed OpenBSD first and had not a lick of trouble. The option I had once I encountered the trouble with disklabel was to go back and change the layout (I combined /var/log back into /var) on the Linux and OpenBSD systems and re-install them. But it just seemed silly to do that if there was a simple workaround in FreeBSD. I was wanting to see if the time it might take for me to get FreeBSD working the way I had planned would be less or greater than the time to go back and redo that Linux and OpenBSD.

    Oddly, as it turns out, there was a way. Either no one knew it, or it got lost in all the noise (a lot of people were transposing the slice and partition concepts, which is easy enough to do since what BSD calls a slice is called a partition in DOS terms). You ask why anyone would want more than 7 partitions in a single slice? and this gets to the nature of the solution, that FreeBSD supports more than one disklabel, each located in different slices. That could have been the solution, but no one presented the whole concept (there were a bunch of things said that made no sense at all, so maybe someone referred to this with an assumption I already knew how FreeBSD worked).

  • I understand your frusteration with your prof. But, as far as I know, you could have released it under GPL for his class and then released it under BSD or whatever afterwards. If you are the copyright holder, I believe that you can do this.


  • The Debian project not only has the "Debian GNU/Linux" distro, they also have "Debian GNU/HURD". I can't think of any reason why the BSD folks couldn't make their own "Debian GNU/BSD" distro, and if they did, I would be willing to run servers with that. Why not? The vast majority of your day-to-day work is with system utilities, most of which come from the GNU project. If I were running BSD, I could still have my tcsh, the arguments to ls wouldn't be any different, and so on. How often do you really care which kernel you are running?

    There can never be a GNU/BSD. BSD's nature as an integrated OS prevents it. The kernel and userland are designed to be integrated. Linux's nature as just a kernel LENDS itself to being packaged with other people's userland utilities (Whence distributions are born as everyone thinks their choice of userland is better.) The Debian developer who first thought up the idea of a GNU/BSD was forced to rethink his position. His misunderstandings about BSD's development were pointed out to him and made him change his mind. Follow the original thread on the matter.

    If ls is ls and tcsh is tcsh, then what does it matter what userland you use? Or did you think tcsh isn't really tcsh on BSD?

    But maybe the BSD folks don't want to do things the Debian way. (For example, I believe the /etc directory in a Debian GNU/Linux system looks very different from /etc in a BSD system. The BSD folks probably like it just the way it is.)

    It's not the "Debian way" but the SysV way. Debian, like just about every other Linux distro (Except Slackware which is BSDish) out there uses a SysV style init. BSD, being BSD, naturally uses the BSD style init. You just have limited Unix experience, which is ok, but your frame of reference suffers for it.

    They still have a chance at winning me over: they just need to code up a BSD version of apt-get. (This implies Debian-style packages... does BSD even have packages?)

    Be careful there, that's almost inflamatory. Of COURSE it does. The much touted Ports Collection is the preferred way to install third party programs, but pre-compiled packages (themselves made from Ports) are available. Does it really matter if you type "apt-get install postfix" or "pkg_add -r postfix" (for example)?


  • What about preserving configuration files?

    That is that the mergemaster step does. Deals with conf files you havn't touched, and gives you diff's for the rest.

    Just to point out that mergemaster is specifically for upgrading FreeBSD's /etc. Third party (ie, "local") apps config files are installed in (typically) /usr/local/etc. Ports/Packages handles upgrades by installing the vendor supplied config files as "foo.conf.default". You would "cp foo.conf.default foo.conf" and edit it. Since "foo.conf" isn't in the Package List, it won't be deleted when you pkg_delete a program before "make install"ing a new version. You would then wind up with a NEW "foo.conf.default" which you would use to merge in any config changes into YOUR "foo.conf".

    Automatic upgrades are not to be trusted IMHO. Software configs _and_ behavior can change from release to release. Blindly running any command that updates software without my knowing ahead of time what I'm installing is asking for trouble.


  • by jabbo ( 860 ) <jabbo@yahoo. c o m> on Tuesday February 06, 2001 @06:22AM (#454809)
    Was that a joke?

    The situations where Linux beats up on FreeBSD are rare. I know. I've seen a couple of them in live production. Most people won't. Ever.

    You are exaggerating the importance of SMP, I think. Yes, 2.4 (and even 2.2) have Real SMP support, as opposed to FreeBSD 4 and earlier's "one processor for kernel space, one for user space" approach. (If that's not the God's honest truth, then it's somewhat pathetic that a quad-CPU box running threaded apps at 4.0 utilization only utilized two processors in the deployment I'm thinking of) Most users won't have to care about this if they're running x86 hardware. I have seen very few sites that needed scalable SMP and couldn't afford a Sun or Sequent box at the heart of their business. (and I simply don't trust Intel hardware for applications where I can only afford to spec a single unit in production; for clustering and server farms, it's fine)

    Moreover, FreeBSD 5's SMP support is likely to be on a par with 2.4's. That's pretty damn good. I haven't seen many commodity 8-way Intel boxes, and I've never seen any non-Sequent 64-way x86 machines. Ever. (I've used plenty of the rest)

    On dual-processor machines, especially webservers, I have watched FreeBSD kick the crap out of Linux. On quads, Linux does indeed beat up on FreeBSD. But most people are not purchasing 4- or 8-way chassis with an eye to expansion, the way people do with Suns (eg. 4500's with a single processor board, etc).

    What worries me is that the same people who consider your post 'informative' are the ones most likely to come to the wrong conclusion.

    Oh well. Anyone who believes everything they read on Slashdot shouldn't be making purchasing decisions anyways...
  • by isaac ( 2852 ) on Monday February 05, 2001 @03:38PM (#454810)
    Why Linux over FreeBSD for servers?

    How about mature, working SMP support?

    I swear, all these comments along the lines of "well, FreeBSD might be better for big servers, but..." comments crack me up. Fact is, Linux (particularly the 2.4 series kernels) trounces FreeBSD when it comes to scaling on high-end (read: SMP) x86 hardware.

    FreeBSD is probably still better for single-cpu boxes (think Hotmail or Yahoo's server farms), though I haven't yet tested FreeBSD against a 2.4 kernel in this configuration.

    Don't take this as a flame against FreeBSD; I cut my teeth on netbsd and 386BSD back in the day. I just find the lack of proper SMP support is a showstopper when it comes to deploying Free/OpenBSD in many environments.


  • by rve ( 4436 ) on Monday February 05, 2001 @03:46PM (#454811)
    What religious war? They're just 3 different OS-es for different situations. FreeBSD sacrifices some portability for better performance and ease of use on the intel PC. NetBSD sacrifices some raw performance and ease of use for the broadest possible portability, and OpenBSD sacrifices some performace, ease of use and new- or exotic hardware support for more guaranteed security.

    It is nothing like windows versus linux, or linux distro A versus linux distro B. When lots of people say they don't like Theo de Raadth, that doesn't make it a holy war.
  • by RelliK ( 4466 ) on Monday February 05, 2001 @02:55PM (#454812)
    He explicitely said that these benchmarks are not scientific and are merely his opinion. Therefore, this error is permissible. Besides, this optimization depends highly on the compiler used. My understanding is that gcc does not do as good a job optimizing as some other compilers. Furthermore, even if this stuff did get optimized out, I don't see how it would change the outcome of this comparison.

    All in all, this is the best comparison between FreeBSD and Linux I've seen yet. Everything else I've heard up until now was basically anecdotal evidence and hearsay. Yes, this is not a real benchmark. But at least it shows some evidence.

    Notice how he says that tweaking Linux improved performance a lot. If anything, this shows that out of the box, FreeBSD is better tweaked for a server than Linux, which makes sence -- FreeBSD is primarily a server OS, while Linux is targeted for both desktop and server, and different distributions configure the kernel differently. What I would like to see is a benchmark of fully-tweaked Linux (once 2.4.x tree stabilizes) and fully-tweaked FreeBSD.
  • by gravity ( 6609 ) on Monday February 05, 2001 @03:08PM (#454813)
    hey man , you should try windows.
    tons of books and lots of sound drivers.
  • by BJH ( 11355 ) on Monday February 05, 2001 @08:40PM (#454814)
    Hardly, why would I have been asking if I didn't already feel that way. My impressions were already "colored".

    In which case, your example of your teacher is rather pointless, since it would seem you had already made up your mind about the GPL - just as he had about the BSD license.

    My point with repeating this story, was that most GPL people are like this.

    This is a ridiculous generalization. Have you talked with a majority of GPL supporters? Or even a small proportion? This is like me saying, "Most BSD supporters are fascists." No supporting arguments, no personal experience - just a blatantly provocative generalization.

    And the GPL itself has the same attitude. You WILL do it our way, or we will find a way to FORCE you to comply.

    That's just ridiculous. If you don't use GPL'd code in your own code, you have no problem. If you DO use GPL'd code in your own code, you should at least repect the license the author of that code put on it.

    Trying to make me look like a childish person throwing a mindless tantrum is a nice tactic -- but it doesn't work in civilized society.

    More than anything, this sentence makes me think of a teenager saying, "I'm not a kid! Don't treat me like one!" Please, grow up and respect the fact that some people like the GPL more than the BSD license, just as you prefer the opposite.

  • by SimplyCosmic ( 15296 ) on Monday February 05, 2001 @03:01PM (#454815) Homepage
    Exactly. You should use the tool that provides all the features you need, along with support options that suite you.

    Linux support and documentation isn't the best out there. OpenBSD, for instance, has some of the best man pages I've come across. And the book, "Building Linux and OpenBSD Firewalls" was more than enough to help me install and configure a nicely tuned daemon firewall box.

    However, not everyone goes about searching for information the same way, which in the end was my main point. For some, answers will come quickly through *BSD channels. For others, a particular Linux site may be everything they'll ever need. When I was playing around with different installs on a spare box, I never could find the information needed to get a particular sound card working, or to adjust this or that feature under FreeBSD, whereas I happened upon the solutions quickly for the current Debian setup I have now.

    So, as I said originally, the main reason I use Linux isn't because of it's technological superiority or even better support, just that the layout of the user support channels happens to suit me better than it might someone else.

  • by cosis ( 23720 ) on Monday February 05, 2001 @10:31PM (#454816) Homepage
    We're sorry! Your request has generated an error of some kind.

    The error has been logged and will be examined promptly by our technical staff. We apologize for the inconvenience.

    Go to Byte home page
  • by seanb ( 27295 ) on Monday February 05, 2001 @07:25PM (#454817) Homepage Journal

    A while ago I was issued a Dell Inspiron 7500. After taking a good hard look at the Linux distros out there (I was running Debian on my desktop at the time, I decided to put FreeBSD on the laptop. These were my major reasons:

    1. USB support. This was before the days of linux 2.4, and FreeBSD USB support was rock-solid. At one point I was using a USB mouse and a USB Ethernet adapter, and both just worked. I ended up playing with the lower details of USB later, but that was for my own curiosity - eveything already worked.
    2. Faster boot times. In my (subjective) experience, FreeBSD booted in about half the time of linux. This is a big deal for me.
    3. The FreeBSD boot loader. I was sick of mucking around with LILO every time I compiled a kernel.
    4. fsck. Working with a laptop, it is almost inevitable that the machine will sometimes be shut down improperly. Again, this is subjective, but I found that this causes problems much less often in FreeBSD, and any problems that are caused are usually automatically fixed by the boot process. This just makes the boot process take much longer, comparable with the start-up time of a linux system.
    5. Curiosity. I had heard how cool cvsup/make world could be, but I was skeptical. Coming from the Debian world, I doubted anything could be better than apt-get, so I wanted to find out.
    6. I was fed up with packages (wrestling with qmail on Debian was the last straw!). FreeBSD works VERY well for people who like compiling from source, and makes this option amazingly convenient.

    Do I think FreeBSD is ideal for everything? No. I'm typing this from a Mandrake box I set up for my roommate to use.

    Have I become a BSD bigot? Possibly. As I type this I wear a FreeBSD shirt and a tattoo of Chuck, the FreeBSD daemon.

  • by powerlord ( 28156 ) on Monday February 05, 2001 @06:50PM (#454818) Journal
    You're right to be angry about the teacher. The choice of license should have been up to you.

    Why can't both licenses exist in the world simultaneously?

    Personally I look at it like this:
    The BSD license gives the code away for anyone to do with as they will, relying on the kindness of others to repay the gift in terms of giving code back to the community and not abusing it.

    The GPL license is more greedy by demanding payment to use the code in other projects. The payment is that you have to put your code under the same license.

    In truth the BSD license is the more 'free' in terms of the coder, and does reap benefits (as one person put it, Apple didn't base Darwin on GPLed code), but I wonder if the greedier way of the GPL is more benificial to allowing the code to propogate.

    (I'm just trying to leave aside the idea of which is better for the user or which is better for the programmer and just looking at it in terms of which is more likely to provide more code back to the general 'pool', but personally I believe both licenses can co-exist and serve different purposes)

  • by Pemdas ( 33265 ) on Monday February 05, 2001 @03:48PM (#454819) Journal

    But that would be contrary to the general /. methodology of Comment First, Read Second if you feel like it and have nothing better to do. :)

    But seriously:

    He explicitely said that these benchmarks are not scientific and are merely his opinion. Therefore, this error is permissible

    May be permissible to you, but I find such basic errors disturbing. It makes me wonder what he considers to be "linux tuning", and how equivalent his setups really were.

    Besides, this optimization depends highly on the compiler used. My understanding is that gcc does not do as good a job optimizing as some other compilers

    GCC gets this one right, regardless of architecture, unless you explicitly tell it to do no optimizations. (-O0). Check it out...compile that program with gcc and objdump -d it; you'll see no FP ops at all.

    Furthermore, even if this stuff did get optimized out, I don't see how it would change the outcome of this comparison.

    It probably just skewed the results more towards process creation/reaping, which is appropriate, given this is supposed to be an OS benchmark. But the fact that he added such pointless code shows he doesn't have a really good grasp of what systems and areas he's stressing (or trying to).

    All in all, this is the best comparison between FreeBSD and Linux I've seen yet. Everything else I've heard up until now was basically anecdotal evidence and hearsay. Yes, this is not a real benchmark. But at least it shows some evidence. It's better than a random "FreeBSD Rulez" comment on usenet or equivalent, but not by much, in my book.

  • by SecretAsianMan ( 45389 ) on Monday February 05, 2001 @08:04PM (#454820) Homepage
    GNU/BSD? Would Debian dump the BSD userland (which is part of what BSD is) and replace it with a totally GNU userland?

    SecretAsianMan (54.5% Slashdot pure)
  • by tmontes ( 80312 ) on Monday February 05, 2001 @02:21PM (#454821)

    Hmmm, there is clear evidence of a typo in the article. I certainly believe they meant...

    'This one is a dual CPU system with PentiumIII 900-GHz processors and 768 GBs of RAM. The disks are under a RAID controller, letting the five 18.2-TB disks be visible under RAID5.'

    ...these boxes wouldn't scale otherwise, would they ?:)
  • by AntiBasic ( 83586 ) on Monday February 05, 2001 @03:30PM (#454822)
    Well, I look at it this way. I can walk into any bookstore and get an O'Reilly book detailing how to write drivers for Linux, another explaining Linux internals in detail, yet more describing for newbies how to install same. There are wonderful distributions like Debian and (well, at least when they can make a release that allows you to get the kernel to compile) RedHat, etc etc etc. On the daemon front, I've seen books available by mail, none in the bookstores. There's certainly a lot less in terms of choice.

    Don't obfuscate quantity for quality. Sure when I go to Borders, I see about 5x the number of books on Linux; but all but two or three of them are on "Debian for Dummies" or "Red Hat Unleashed". They aren't really technical. I can't read Debian for Dummies and understand how Linux handles those new fangled zero-copy sockets. Now there are the couple of Linux kernel books but there is also the 4.4BSD red book. It explains the whole OS! Not just the kernel. If you've ever read the needs to be updated Complete FreeBSD by Greg Lehey it's like a happy medium between r33t k3rn37 d00dz and "How to tie your shoes the Linux way."

    Popularity is really not a good reason to choose something. Windows is a lot more popular than Linux. It has more users, more commercial programs, more programmers, certifications, and possibly books, training courses, and any number of other things. It doesn't make it any better, really, now does it? Yes, there are more Linux users than FreeBSD users. It doesn't really make that much of a difference.

  • by elainerd ( 94528 ) on Monday February 05, 2001 @02:46PM (#454823) Homepage
    FreeBSD already emulates most Linux software out there. It runs my SBLive natively and perfectly. I didn't install it because I wanted it to do everything Windows does, I installed it because I wanted a kick-ass Unix system to run at home. I use it for a workstation at home and have it on servers at work. It is rock solid. Easy to Upgrade? Yes every morning it cvsup's that nights changes, fixes, to the src code. Then every month I run make buildworld and make installworld and I have the latest FreeBSD every bin freshened. I used Linux since 1993, since changing to FreeBSD I've had no desire to go back. The people who are complaining about lack of documentation and resources are being silly. Besides the brand name FreeBSD books out there, there are many resources and websites with step-by-step help for newbies. IRC especially has very knowledgeable people.
  • by festers ( 106163 ) on Monday February 05, 2001 @05:45PM (#454824) Journal
    Saying "my benchmark methods are crappy" doesn't excuse you from still posting crappy benchmarks.

  • by mojo-raisin ( 223411 ) on Monday February 05, 2001 @06:18PM (#454825)
    Quite correct.

    According to

    --- is an Intel architecture PC machine running the FreeBSD
    operating system.

    Its configuration is as follows:

    Micron NetFRAME 9201 system, consisting of:

    One 500MHz Intel Pentium-III Xeon CPU w/512K L2 cache
    4GB of main memory (16 * 256MB 50ns ECC EDO DIMMs)
    1 Adaptec AHA-2940U2W PCI single-channel wide Ultra-2 SCSI controller
    2 Adaptec AHA-3940AUW PCI dual-channel wide UltraSCSI controller
    1 Intel Pro/100+ PCI 100Mbps Fast Ethernet controller
    1 Bay Networks Netgear GA620 Gigabit Ethernet adapter

    Please visit for more
    information on the NetFRAME 9201 system.


    That micronpc link seems to be busted though.

    A picture of the beast is here:
  • by raju1kabir ( 251972 ) on Monday February 05, 2001 @02:10PM (#454826) Homepage

    Okay, well, this only makes sense if you have several machines (I've never seen the appeal of dual-booting unix), but there are definitely comparative advantages to each OS.

    For mail servers, DHCP, DNS, NFS, firewall, NAT, etc., FreeBSD means less headaches. I've got plenty of FreeBSD boxes that have never been down except for OS upgrades or hardware moves. You can lock them in a closet and never think about it again - it's like the golden days of Novell Netware all over again. And under intense loads, say, a well-utilized IMAP or Samba server, FreeBSD keeps its chin up far longer than Linux.

    However, the problem with FreeBSD is that, let's face it, the userland is just not as friendly as that in Linux. You spend days installing all the happy-fun GNU tools with useful command-line arguments and post-1970 approaches to system management, and by the time you've done that, you've kluged together a system only a mother could love. So, for day-to-day messing around, Linux can be much friendlier.

    Perhaps even more importantly, the new generation of commercial software is largely just not available for FreeBSD. Want to run Oracle? Domino? You can try to shoehorn them into FreeBSD's Linux emulation environment, but I can tell you from painful experience, it's not pretty - if it works at all. And when you try to do things like linking other third-party software against the Oracle libraries under Linux emulation, you'll spend weeks up to your eyeballs in Makefiles and header files. Not worth the trouble, even for the incremental improvement in stability.

    So they both have their places, and they're both well-worth learning about. But I'd be quite suspicious of someone who claims that one is a "better" OS than the other - it depends far too much on one's specific needs.

  • by geomcbay ( 263540 ) on Monday February 05, 2001 @03:57PM (#454827)

    The errors call into question his overall credibility as a 'guru'.

    He states that he has introduced floating point into the C program (to make it a tad 'tougher'), when he hasn't even done that! He's trying to asssign a floating point number to a long (which is an extended int, maybe he meant double?) so the compiler is just going to chop that number to a (long) int for him, hopefully at least giving him a warning which I guess he ignored. This guy holds a masters degree in CS?

    Nobody is expecting him to come up with a perfect benchmark, but 1) he could at least know something about what he's talking about before he talks about it and 2) there's plenty of existing benchmarks for web performance that he could have used rather than creating his own questionable benchmark.

    For what its worth, I run FreeBSD, not Linux, so it's not that I disagree with his results, just the way they were obtained.

  • by izzertaq ( 304006 ) on Monday February 05, 2001 @02:28PM (#454828)
    The article said something about Linux performing much better after hand-tuning the VM, which begs the question, seeing as how FreeBSD tunes its own VM for good default settings, why can't Linux do the same? This is just like the IDE DMA situation on BSD vs. Linux -- FreeBSD has had autodma working forever on VIA chipsets, and Linux, even in 2.4, defaults to PIO mode on IDE disks unless you enable 'highly dangerous' code. Oddly, the I/O elevator in 2.2 has a sensible out-of-the-box setting, but 2.4 requires tuning with an arcane tool called 'elvtune' or something if you want an elevator at all. Seems like a step backwards to me, regardless of how technically superior the 2.4 way is.

    Unix purists go on about how the kernel should never set policy, but that's rather silly. Really, the kernel-'enforced' policy is whatever the defaults are, as most users expect the default behavior to be intelligent. People install Linux all the time and complain about how they can't do anything without getting choppy audio and mouse movement, because IDE defaults to PIO. It's rather sad that people get bad impressions of Linux because of its braindead default settings for so many things, when it's capable of doing much more.

    Maybe this is work for the distros to be doing, I don't know. I suspect most of them would prefer to pass the buck as well.
  • by dvdeug ( 5033 ) <dvdeug@e m a i> on Monday February 05, 2001 @02:37PM (#454829)
    It shouldn't be a matter of dumping the Hurd - if Debian dumped the Hurd, we'd lose the developers that develop the Hurd. But if you want Debian GNU/BSD, subscribe to and start working. The biggest problem with Debian GNU/BSD is interested workers, not anything political and certainly not anything having to do with the Hurd.
  • by Cerb ( 10299 ) on Monday February 05, 2001 @02:19PM (#454830) Homepage
    The FreeBSD handbook is pretty much THE reference for FreeBSD, and it's on every FreeBSD box that has the docs installed. You can get a dead tree version too if you want. I know of at least one other dead tree text on FreeBSD, it has regular updates posted to the list.

    And what's this about no sound drivers? When was the last time you actually used a FreeBSD machine? Or, if you claim to be a FreeBSD user, have you not read LINT? There are tons of sound drivers. The pcm device runs many PCI and ISA cards. And what qualifies as "and the like"? My hauppauge WinTV card works better in FreeBSD than it did in Linux. The machine doesn't slow down when I watch TV. My USB mouse was spotted and configured as soon as I started the install. The ONLY thing I miss about Linux is the Nvidia drivers.

    No, I'm not a FreeBSD bigot. I use both Linux and FreeBSD. If you look, I'm actually a Debian developer. They both have a place in the world. But, if you are going to post something, post facts.

  • by jackal! ( 88105 ) on Monday February 05, 2001 @02:06PM (#454831) Homepage
    We should all switch to FreeBSD anyway now that Microsoft has declared Linux doomed!


  • by ssimpson ( 133662 ) <slashdot&samsimpson,com> on Monday February 05, 2001 @02:08PM (#454832) Homepage

    From the article []:

    'I chose -- once again -- IBM's Netfinity 5100 server. This one is a dual CPU system with PentiumIII 900-MHz processors and 768 GBs of RAM. The disks are under a RAID controller, letting the five 18.2-GB disks be visible under RAID5.'

    Damn. Makes my half a gig of RAM look very sad :)

  • by popular ( 301484 ) on Monday February 05, 2001 @02:07PM (#454833) Homepage
    OK, no hitting below the belt.
    Let's see a good clean fight.

    Also on tonight's fight card:
    GNOME vs. KDE
    Perl vs. PHP
    MySQL vs. Postgres
    RMS vs. ESR


  • by frob2600 ( 309047 ) on Monday February 05, 2001 @05:39PM (#454834)
    I also don't want to see someone else taking my work, "embracing and extending" upon it, and then profiting off a piece of software that I originally intended to be free for everyone. The GPL guarantees that what I intended to be free will stay free.

    Don't you see? The code you have written and intended to be free is still free! Nothing changed that. If they have added, or improved your code, so much -- they should have the right to do what they wish with that code. They are only benefiting from their work. They can never really benefit from yours. Why? It may appear like they are benefiting from the work you have done, but they are not. They can never force people to buy their code on the merits of your code alone. Because the people will just go out and use your code; if they don't want the additional stuff.* What gives you the right to tell them what they can or can't do with the code they wrote? In a truely free society, nothing.

    I can see where the GPL people come from with their ideas. And the ideas are noble. But you are going about it in the wrong way. Be honest with yourself and read the GPL (I have a nice highlighted hard copy for when I discuss these issues). Ask yourself if I have a point. If I do have a point, think about what you are doing when you GPL a piece of code. I am not going to tell you that you have to switch or you are anti-anything. I will just ask you to be honest with yourself.

    Personally, I used to think the GPL was the only way to go. I didn't know much about other licenses out there, at the time, and I would have probably argued the same way as many of you do. But, over time, I started to question the ethics of forcing someone to chose one license over another and that in itself is so against freedom that I had to change my view. I am not saying the BSD license in the best [although it is my favorite]. You might want to release code under the Beer-Ware license.** I do this with code I write for people on campus and it pays off very well.

    *Yes, I know they are able to just sell your code without changing anything. But, they will almost never succeed at locking everyone into using their code that way. Because people will still be able to get yours for free. Remember, you have chosen what gets done with the code you wrote. Nothing, and I mean nothing, will change that unless you do it yourself (and even then -- good luck getting that opensource code locked down).

    **Note: I did not write this, I just like using it.

    "THE BEER-WARE LICENSE" (Revision 42):
    wrote this file. As long as you retain this notice you
    can do whatever you want with this stuff. If we meet some day, and you think
    this stuff is worth it, you can buy me a beer in return. Authors Name

    "Do not meddle in the affairs of sysadmins,

  • by stripes ( 3681 ) on Monday February 05, 2001 @02:17PM (#454835) Homepage Journal
    FreeBSD is a pure server OS. Nobody has to worry about the other possible applications, it is designed purely for one purpose, and one purpose alone. It does it well.

    If FreeBSD is a server-only OS why did it get USB support before Linux did?

    If FreeBSD is a server-only OS, what is PicoBSD all about?

    Fact of the matter is FreeBSD serves multiple intrests as well. And it does them all reasonably well.

  • by SimplyCosmic ( 15296 ) on Monday February 05, 2001 @02:52PM (#454836) Homepage
    Hey, although it's an attempt at humor, your post is actually kind of close to my philosophy of "using the right tool for the right job".

    I've had plenty of people ask me if they should switch from MS Windows over to Linux. For some the answer is yes. But only if they're ready for the headaches that come with breaking from the pack. For others, I realise early on that they're going to be much more productive on a machine with an operating system that they're pretty much guarenteed to be able to be fixed by the local Best Buy.

    It really isn't about which OS is the holy grail, perfect for all situations and godsend to all who use it. That's because such a beast doesn't exist. It's about finding the right tool for the job, and the right tool means not only the proper amount of control and features, but support and comfort for the person using it.

    Despite what we geeks tell ourselves, an operating system is just a tool, not a lifestyle. Right? Right? Guys? Hello?

  • by SimplyCosmic ( 15296 ) on Monday February 05, 2001 @02:07PM (#454837) Homepage
    I've played around with several of the distros, as well as FreeBSD and OpenBSD, and the absolute most basic reason I use Linux over FreeBSD is simply because I have an easier time finding out how to install, maintain, admin or fix some problem from the various Linux sites out there.

    It's not really a matter of which is technologically superior, and I suspect that FreeBSD may in fact be so. However, in the particular style of searching for information on how to accomplish a particular task, I've always found the Linux information quicker and easier than for the FreeBSD way of doing things. Again, this doesn't mean that Linux is better, far from it. It's just easier for me to run thanks to the types of online resources I come across.

    Your mileage, as always, may vary. Offer void in most major cities. Not to be taken internally, while pregnant, or running for Congress.

  • by Pemdas ( 33265 ) on Monday February 05, 2001 @02:34PM (#454838) Journal
    The overall conclusions may be valid; I don't know. I've used NetBSD for a few things here and there, but don't have enough experience with FreeBSD to make any sort of judgement.

    This, however, caught my eye:

    • int x;
      long y;
      y = 28.2839281;
      x = 339829;
      y = x / y;

    Notice how I included some simple floating point arithmetic in the C program to make things just a tad tougher.

    He admits he's no benchmark specialist, but any compiler worth its salt (and many that aren't) will optimize the floating point operations away. Also, since the result of the divide is never used, that will be optimized out, too.

    I don't know what the real story is, and I do know a lot of knowledgeable people split on the Linux vs. FreeBSD issue. However, such a blatant error in benchmarking methodology gives me large doubts about this guy's credibility as a competent judge.

  • by AntiBasic ( 83586 ) on Monday February 05, 2001 @02:58PM (#454839)
    And because when I talk to the network managers at work, my fellow consultants/contractors, and my clients, they all talk about Linux, not FreeBSD. I've convinced a few of the wonders of OpenBSD and audited source, but many still compare Linux and FreeBSD in terms of market share: who is the bigger? Linux. Which is a customer more likely to ask for? Linux over BSD, but Solaris and AIX and HP-UX above Linux.

    Popularity is really not a good reason to choose something. Windows is a lot more popular than Linux. It has more users, more commercial programs, more programmers, certifications, and possibly books, training courses, and any number of other things. It doesn't make it any better, really, now does it? Yes, there are more Linux users than FreeBSD users. It doesn't really make that much of a difference.

    And besides that, FreeBSD out of the box isn't as friendly as most Linux distributions.

    Care to justify that statement? Easier to learn, again, is questionable. It's easy to learn something if you have, say, a friend next door that runs the same thing. At my university, FreeBSD became very popular (much more so than Linux) because the people who took the time to help out and organize things knew FreeBSD best, and suggested people try it. Those same people who used Linux before considered FreeBSD much easier to learn. The same may apply the other way around in your area. It isn't a matter of ease, but your surroundings. If you go it alone, like I pretty much did, it ends up being a personal matter (discussed below). As for documentation, I'd say it depends on the person. The FreeBSD Handbook helped me through most of my trials, but some find it too complicated, and some find it too abstract. Greg Lehey's book is good. There're FreeBSD courses offered by BSDi, amongst others. The NetBSD documentation is technically great and complete.

    Maybe when my hardware needs change, I'll run FreeBSD. If FreeBSD NFS and Linux NFS start talking to each other faster,...

    The problem there is Linux NFSv3 implementation. Quite honestly it sucks.

  • by Pinball Wizard ( 161942 ) on Monday February 05, 2001 @02:42PM (#454840) Homepage Journal
    I realize you were joking but... MS would never say nasty things about the BSD's since their TCP/IP stack and kerberos are largely based on BSD code.

    They love BSD for this reason. They have told their developers to not even look at GPL code while on the job.

    If anything BSD is doomed by their license(trying hard not to troll here). Most open source developers would rather not have their code end up in Windows. Hence, eventually the popularity of Linux and the GPL with developers will mean that Linux will likely overtake the BSDs in performance in the not-too-distant future.

  • by bob x johnson ( 197558 ) on Monday February 05, 2001 @05:09PM (#454841)
    Yeah, MS may have used *BSD code for some of their TCP/IP utils, and great code it may be, but has *BSD gotten anything in return?

    You mean besides the benefits of standard protocols?

    [In case you fail to fail to miss the point, like most mindless GNU drones, answer this question: How many successful protocols have their reference implementations licensed under the GPL?]

  • by BluedemonX ( 198949 ) on Monday February 05, 2001 @02:08PM (#454842)
    Well, I look at it this way. I can walk into any bookstore and get an O'Reilly book detailing how to write drivers for Linux, another explaining Linux internals in detail, yet more describing for newbies how to install same. There are wonderful distributions like Debian and (well, at least when they can make a release that allows you to get the kernel to compile) RedHat, etc etc etc.

    On the daemon front, I've seen books available by mail, none in the bookstores. There's certainly a lot less in terms of choice. And you can forget finding sound drivers, or the like. What you do get is the suggestion to take the drivers from the Linux people and port them yourself. Of course you can do that, because you're a g0d l337 ha>0r d00d, right? Otherwise you'd be running Windows.
  • Linux is being pushed in several different directions by different groups and organisations. Some want it to be a Desktop OS, some wish it to be a Server OS, and some wish it to be in the world of embedded devices. Everybody has a different agenda for a free OS, and the means, if they wish, to take Linux and mould it for their wishes.

    FreeBSD is a pure server OS. Nobody has to worry about the other possible applications, it is designed purely for one purpose, and one purpose alone. It does it well.

    If I were running a server alone, I would use FreeBSD. For any other purpose, I would use Linux. Each have their strengths.

    They fuck you up, your mum and dad.

  • by frob2600 ( 309047 ) on Monday February 05, 2001 @03:40PM (#454844)
    Most open source developers would rather not have their code end up in Windows.

    Being an open-source developer myself, I would personally never release any of my code under the GPL. And I encourage all the coders I talk with to do the same. And once they see the reasoning behind why, most will switch. They honestly want to give their code away for others to use as they see fit. You are not giving your code away if you are requiring something back.

    I tried to use the BSD license in a class where the teacher was pro-GPL and he refused to agree. Although we discussed it at great length. It turns out that at my college any professor has the right to determine the license for all code his class produces. But, even if the school had not been on his side the GPL would have allowed him to win. All he had to do was require that a GPL'd base class be included in all assignments, and everything would be forced under the GPL! He told me he was going to do this, no matter what verdict he school returned. This is just one of the reasons I hate the GPL.

    I am sorry, but your little disclaimer about trying not to troll is not going to change what you said into anything less offensive to those who have been forced into submission by that filthy beast of a license! If the cost of giving my code away means it goes to Windows. Great! This again proves my favorite saying, "Those who hate Windows use Linux; those who love Unix use FreeBSD."

    If the GPL becomes as prominent as you think it will be, there will be a lot of excellent coders who will take up other hobbies where freedom is maintained. But it won't happen. Communism failed in use with politics it will fail in use with code developing as well. FreeBSD will remain 'Free' and so will all the code I, and those who know the truth about giving, produce.


    "Do not meddle in the affairs of sysadmins,

Nothing is finished until the paperwork is done.