Organizations that want to use a public Unix variant have two solutions from which to chose: Linux and BSD. The much talked about Linux camp contains a variety of distributions that include different utilities and tool sets. The same is true of the less frequently covered BSD camp. This article compares and contrasts the four main BSD variants and offers recommendations for both server/desktop solutions.
Differentiating Among BSD Distros
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
This article tells the same as all the others: FreeBSD is more performant, NetBSD is more portable and OpenBSD is more secure. Nothing new.
how is it to have freebsd on the desktop?
do they have kde’s packs that new? like kde rc2 ?
do they only have src? or?
I run FreeBSD 5.2.1 as a desktop, but with Gnome instead of KDE. I have all apps that I want with only one “make install clean”. I personally prefer Gnome but KDE runs well as wel. It’s nice to have a desktop with a good backend 🙂
I have FreeBSD 4.10 running on my laptop as a desktop system (with IceWM), and it works just fine. While the base system may be somewhat conservative (and that’s a good thing), the Ports tree is very up to date, e.g. KDE 3.2.3, Firefox 0.9.3, Gaim 0.81, etc…
There are packages for most things on fbsd. The newest cutting edge kde rc2, maybe, maybe not. Since release candidates aren’t designed to be around for very long, there might not be a package. You can try http://freebsd.kde.org/ if you want. I know there is a link to a page of packages built from cvs but I do not know how up to date they are. You could try a web search too, somebody might have built one.
Basically, if it’s something temporary like a release candidate, that only has a couple week lifespan, don’t expect to see a packages. An official release – there will be a package. Considering the fact that most poeople that try out release candidates will build from source, and there are few binary packages of them available anyway, I don’t think this has much bearing on choosing to go with fbsd or not.
i dislike the old myths…
people say “slackware is known for its stability and speed”. and they keep saying it – but they never back it up. i’ve seen no figures. and my own unscientific tests show that other kernels from other distros performs faster. and stability – how do they test that. i’ve not seen people stress testing a line of linux distros to compare? “stability” once associated becomes hard to lose …
same with the BSDs … people are too quick to say that freebsd is mroe performant that netbsd. that may have been true 4 years ago.
same with the BSDs … people are too quick to say that freebsd is mroe performant that netbsd. that may have been true 4 years ago.
Fat chance of that happening, especially when half the myths are started by BSD zealots. For example the most infamous one is “FreeBSD is more scalable than Linux”, which has never been true. “FreeBSD has a better network stack than Linux” hasn’t been true for probably nearly 10 years, although that still persists too
Hubert Feyrer’s (biased 🙂 recommendations:
http://www.feyrer.de/NetBSD/nb2002-lu.pdf
* FreeBSD for high performance SMP based PC servers
* OpenBSD for crypto hardware drivers or filtering bridge support
* NetBSD for everything else 🙂
the last scalabiltiy benchmark i saw what quite amusing.
linux was first.
freebsd was second , but with a comment that usb is working for 2 years and was no dirty hack as in linux
then there was netbsd quite close to freebsd
and openbsd was the last, quite unsuable , besides you only want a firewall..
maybe ill find the link, the article misses DragonflyBSD which may become very interesting soon
Well,
This article started out OK, despite the fact that about half way through the NetBSD secton he called NetBSD FreeBSD…I thought news was checked by editors? Also, once he started talking about OS X…woah boy… talk about pushing ^_^ and he wasn’t even talking about *BSD there mostly, he was touting its availoable software…which ISN’T available for people who simply use Darwin(MS Office, Adobe *, etc…)
Oyasumi Nasai,
–Luzer
Is also based on MACH and there is no mention of it whatsoever.
Also http://www.opendarwin.org is a good place for information on darwin. A lot more alive than apple’s opensource pages.
Ludovic
—
http://perso.hirlimann.net/~ludo/blog/
[quote]
same with the BSDs … people are too quick to say that freebsd is mroe performant that netbsd. that may have been true 4 years ago
[/quote]
http://bulk.fefe.de/scalability/ <– First READ then BLA!
Hello all,
Its interesting to see that people click the “Report abuse” button because they dont like the opinions of others. No one thus far has said anything offensive. And I guess that would be my opinion, I hope no one gets offended. All *BSD’s have their place within the computing universe.
As for FreeBSD on the desktop and available software, please check out:
http://www.freebsd.org/ports
NetBSD 1.6.1 was very old release compared to others which were benchmarked. However, FreeBSD has more developers (I think), so it is possible, that in many ways it performs better. Someone has to do new benchmark with NetBSD current.
I could get the same information just reading the Home Pages of the respective distributions. This article seems to base it’s decision on marketing
FreeBSD = Server/Desktop/Web
OpenBSD = Exposed Internet Server
NetBSD = Portable Applications
Ugh…
Please Note:
A REAL FACT.
Linux kernel has been improved by companies using copyrighted material and literally stealing codes from many commercial UNIXs and even BSDs. ex. BSD Packet Filter.
Coding way of Linux Kernel:
Really this is how linux started.
copy every thing from BSDs or some other UNIXs and rewrite everything using different names.
Its IMPOSSIBLE to code from scratch like Linux.
its just a simple bunch developed from stolen codes.
(ex. BSD Packet Filter codes which were found while searching for SCO UNIX codes in Linux Kernel, if BSD codes are used, then how can anyone call Linux is developed from scratch. ha ha ha, funny isnt it…)
BSDs are the ones which were during the starting time of UNIXs. so be pride to use BSDs !!!
Read BSD License
How can Linux steal code from someone that gives it away freely? I think you may want to do somemore reading on this topic.
Oh and stop contributing to religious wars!
read plain text properly.
from: http://bulk.fefe.de/scalability/
[Nov 1 2003] I got an email suggesting that I re-check NetBSD. The results are nothing short of astonishing. In two weeks time the NetBSD team made dramatic improvements.
* socket: previously O(n), now O(1).
* bind: greatly improved, but still O(n). Much less steep, though.
* fork: a modest O(n) for dynamically linked programs, O(1) for statically linked.
* mmap: a bad O(n) before, now O(1) with a small O(n) shadow.
* touch after mmap: a bad strange graph in 1.6.1, a modest O(n) a week ago, now O(1).
* http request latency: previously O(n), now O(1).
Congratulations, NetBSD! NetBSD now has better scalability than FreeBSD.
PS:NetBSD just rocks.
Let’s see… on one hand SGI is looking at 1024 and even 2048 CPU single system images (not clusters) with Linux 2.6 (see http://www.computerworld.com.au/index.php/id;1463359743;fp;16;fpid;….. and http://www.theinquirer.net/?article=17491 last paragraph).
On the other, FreeBSD 5 had been recently found to be unable to scale a simple MySQL test to 2 CPUs http://marc.theaimsgroup.com/?l=freebsd-amd64&m=108490922809652… – and they were even Opteron CPUs, which have a pretty scalable multiprocessor architecture.
OK, so this says all there is to say about Linux/FreeBSD scalability? You’re right, these are just facts, and there is no reason for a FreeBSD user to get upset over them. Of course, it is always easy to put together two or three “facts” and then present them as a complete argument.
But in fact, there are many aspects of scalability, and the two you mention above are very extreme examples. On the one hand you have a heavily patched Linux kernel by a company that (hey!!) specializes in large SSI clusters. Oh, I’m sure there’s no way they could do the same thing with FreeBSD. And, on the other hand, you have someone in a newsgroup talking about an unstable, in-development kernel (which defaults to having all sorts of debugging extras turned on, which are a performance hog), AND, they are running this on a new architecture (Athlon64), which is still not officially supported by FreeBSD. Not to mention that it is only MySQL we are talking about here, not all server software. MySQL is known to have scalability problems with FreeBSD threading, and all MySQL development tends to be aimed at Linux anyway. Try the same kinds of tests with PostgreSQL, for example, or Apache, on a kernel. As shown in the last serious scalability tests (http://bulk.fefe.de/scalability/), Linux 2.6 and FreeBSD 5.2 (and the updated version of NetBSD) were not that far apart, even though 5.2 is an unstable kernel. IMHO, Linux and FreeBSD are both doing some great things. I hope DragonflyBSD carries that on. By the way, one of the main aims of DragonflyBSD is to handle large-scale SSI clustering. It’s going to be interesting ;-).
Please, this is not a flame, just an observation that one take with a grain of salt the suggestion that the anonymous poster is trying to make.
“Unlike other BSD variants, and most Unix flavors, OpenBSD is installed with everything but the absolutely essential services disabled.”
This is not entirely true NetBSD is the same way.
“OpenBSD is the obvious choice on the server because of its very high security principles.”
Ignorant statement really, I am sure OpenBSD is very secure however few security measures compair to “Security by Obscurity”. Case in point my NetBSD webserver runs on an DEC Alpha PWS. Not to say that a Black hat could not get in but it makes it a serious pain in the butt to compile platform specific exploits. Which usually renders the effort not worthwhile. Even a PS2 running Linux with apache is inherently more secure than a compairable x86 OpenBSD box.
even if the article says about the same any other BSD’s comparison, it’s always fun read the compliments openBSD always get on security, netBSD for portatility and freeBSD for it’s performance on x86 hardware
Let’s see… on one hand SGI is looking at 1024 and even 2048 CPU single system images (not clusters) with Linux 2.6 (see http://www.computerworld.com.au/index.php/id;1463359743;fp;16;fpid;……. and http://www.theinquirer.net/?article=17491 last paragraph).
On the other, FreeBSD 5 had been recently found to be unable to scale a simple MySQL test to 2 CPUs http://marc.theaimsgroup.com/?l=freebsd-amd64&m=108490922809652……. – and they were even Opteron CPUs, which have a pretty scalable multiprocessor architecture.
Then look at these follow-ups…
http://marc.theaimsgroup.com/?l=freebsd-amd64&m=108492283629538&w=2
http://marc.theaimsgroup.com/?l=freebsd-amd64&m=108492662812718&w=2
Its not a fair comparision between them.
In my humble opinion people stare far too much at the technical data when it comes to the choice of system.
Neglected is, what I call the sociological factor.
I personally choose FreeBSD over Linux and NetBSD and whatever, because I met some very competent FreeBSD users and developers and simply like to hang out with that crowd.
Other people might have a stronger affinity to the security paranoids 🙂 over at OpenBSD and no matter if FreeBSD has 5 times increased security in some area, will nonetheless continue to stick with their OpenBSD buddies.
This will become more important, when systems get more mature.
E.g. where do you prefer to live? Italy? Greece? France?
All very nice places.
In the end it is a matter of taste and what people you want to stick with (or don’t want to stick with) what you choose.
Freedom of choice is great by the way.
Regards,
Marc
This article wasn’t bad. If you are new to *BSD then this article is for you.
OK, so this says all there is to say about Linux/FreeBSD scalability? You’re right, these are just facts, and there is no reason for a FreeBSD user to get upset over them. Of course, it is always easy to put together two or three “facts” and then present them as a complete argument.
Well it appears that you find it easier to ignore the facts.
But in fact, there are many aspects of scalability, and the two you mention above are very extreme examples.
I was talking about SMP scalability – I thought that was obvious from the context. There are not many aspects of SMP scalability: for a given workload, the kernel has an integer value representing its scalability from 1 to n CPUs.
On the one hand you have a heavily patched Linux kernel by a company that (hey!!) specializes in large SSI clusters.
They are not SSI clusters. They are cache coherent global shared memory multiprocessors that I’m talking about. They do also do clusters with these systems, but that is another matter.
But the biggest thing is that their >512 CPU systems won’t be done with their heavily patched Linux 2.4, but a nearly vanilla 2.6 kernel. They will have some changes no doubt, but I don’t think much in the way of scalability work should be required.
Oh, I’m sure there’s no way they could do the same thing with FreeBSD.
You’re right.
[i]And, on the other hand, you have someone in a newsgroup talking about an unstable, in-development kernel
They are aiming for a release in a couple of months. So they should be in feature freeze – nearly in code freeze. Although perhaps FreeBSD’s holier than thou “RE” standards are slipping.
(which defaults to having all sorts of debugging extras turned on, which are a performance hog),
If you follow the thread, you’ll find that all debugging extras were turned off. The problem was put to many freebsd mailing lists – even a $500 reward and an offer of access to the problematic systems was given. Nobody could solve it or even tell them what the problem was.
AND, they are running this on a new architecture (Athlon64), which is still not officially supported by FreeBSD.
Really? What does “Tier 1” mean then? http://www.freebsd.org/platforms/amd64.html
Not to mention that it is only MySQL we are talking about here, not all server software.
Linux was able to handle it.
MySQL is known to have scalability problems with FreeBSD threading, and all MySQL development tends to be aimed at Linux anyway.
I agree MySQL is crap, but they generally will develop towards a UNIX (ie. SuS, POSIX) interface. FreeBSD obviously can’t handle it.
Try the same kinds of tests with PostgreSQL, for example, or Apache, on a kernel. As shown in the last serious scalability tests (http://bulk.fefe.de/scalability/), Linux 2.6 and FreeBSD 5.2 (and the updated version of NetBSD) were not that far apart, even though 5.2 is an unstable kernel.
Well, I’m talking about SMP scalability, not algorithmic scalability… but even so, Linux 2.6 hands down beats FreeBSD there too – if not in big-Oh complexity then definitely in absolute numbers.
IMHO, Linux and FreeBSD are both doing some great things. I hope DragonflyBSD carries that on. By the way, one of the main aims of DragonflyBSD is to handle large-scale SSI clustering. It’s going to be interesting ;-).
Whether or not they *can* handle it is another matter. It is not a trivial task.
Please, this is not a flame, just an observation that one take with a grain of salt the suggestion that the anonymous poster is trying to make.
Neither is this
As usual, whenever there is a BSD thread, it goes down the “Linux is better than BSD” road.
Could somebody tell me how long does it usually take for him to update from ports gnome, kde (core) or mozilla?
Depends on wether you are using source or packages. If packages, not long. If source, it really depends on what processor you have.
I mean using source. Does ports upgrade
with precompiled packages? How long does it take with your processor?
You dont see anything because we slackers have nothing to complain about and we have nothing to prove. Either you slack or not. I love Slackware with every Cell in my body, but FreeBSD? Damn its nice…. I love the hell out of pkg_add.
Once you go slack you never go back.
-nX
As shown in the last serious scalability tests (http://bulk.fefe.de/scalability/), Linux 2.6 and FreeBSD 5.2 (and the updated version of NetBSD) were not that far apart, even though 5.2 is an unstable kernel.
There is scalability, and there is scalability. Fefe’s benchmarks does NOT test the different kernels ability do things in parallel (like SMP and multithreading), but algorithmic scalability (had the issue been parallel scalability, there is no way NetBSD would have been able to improve significantly in a such a short period of time). Stop conflating the two.
Parallel scalability is an area in which, right now, Linux is well ahead of the pack.
Depends, you can upgrade ports with packages if you wish.
Firefox takes a little under an hour from source. From packages, it would just be the time it takes to untar it and register it, so not long what so ever.