Home > FreeBSD > Roadmap for FreeBSD 5.3 Roadmap for FreeBSD 5.3 Eugenia Loli 2003-12-25 FreeBSD 100 Comments FreeBSD Release Engineering Team’s Scott Long has outlined future plans for FreeBSD 5.3 now that FreeBSD 5.2-RC2 is getting close to release quality. About The Author Eugenia Loli Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker. Follow me on Twitter @EugeniaLoli 100 Comments 2003-12-25 2:24 am All your complaints are why 4.9 is listed as stable, and these are the issues that Ian wants to have remedied before 5.x becomes stable. I don’t think it’s fair to say that FreeBSD is falling ‘further and further behind’ given that, until just recently, Linux-stable was much worse than FreeBSD stable. FreeBSD has been stabler, faster, and generally more robust for a decade. Only with the 2.6 Linux kernel are we beginning to see a genuine, bonified market leader in many kernel areas. IMHO the *BSDs are held back by their license. Because BSD code can be used in anything (indeed, Microsoft make use of several chunks of BSD code) people are often implicitly reluctant to be contributing directly to the fiscal success of others. Linux, on the other hand, is protected from such violations by the GPL. People could argue the philosophies and the correctness of the different licenses, but at least with Linux you know where your contributions goes and stays: in the public domain. Ergo more companies are willing to hire people to code for Linux in order for them to produce businesses around it. That’s my understanding. 2003-12-25 2:36 am “Linux, on the other hand, is protected from such violations by the GPL.” The BSDL is like a gift. The fact that Microsoft has used it doesn’t make it a violation. Damn it I wish that people would figure that out. The fact that Microsoft is using BSD code DOES NOT MAKE BSD LESS FREE! 2003-12-25 2:38 am You’ve been told before. Go away. 2003-12-25 2:42 am If Linux is *that* good, why do you keep posting in BSD threads? Anyway, you seem too thick to realise that Linux is developing faster because it has far more developers. I also have to agree with Charlie. I’ve used Linux 2.2 and 2.4 quite extensively and it wasn’t what I call really stable. The FreeBSD team decided to change the lock in a single version because they do care of architectural changes. They don’t radical in a so-called stable branch. Like it or not, both OSes are good. What I’m really interested to see is if Linux will sink like many new technologies introduced in the last decade. They got a really high spike… and then people/investors lost interest when they found out it was overhyped. It has happened with the Internet, it might happen with Linux. I’m not saying it will… but it might. 2003-12-25 2:44 am I should preview my messages before posting them. They don’t make radical changes in a so-called stable branch (hint: Linux 2.4 + VM). I don’t hate Linux… Hell, I’m a Linux user and a BSD noob. Both projects are just following different philosophies. 2003-12-25 5:24 am The O(n) ULE scheduler, which is supposedly better than the current scheduler FreeBSD is using, still isn’t finished. It is already better than current one in my case test so far on x86 and I have been using ULE as default on several machines. 2003-12-25 5:45 am And don’t know why people get offended when comparing and criticizing flaws, faults or improper design methodologies in operating systems. It used to be that Linux sucked, because it was young and immature. And that the BSDs when as solid as a rock and unquestionably more mature. In the minds of die-hard BSD users, Linux will continue to be a toy, regardless of how robust it is today. Just like in the minds of your parents, you’ll continue to be a kid, even when you have grandchildren. Well, things change, and so has Linux, at a shocking rate I might add. What it took other Unix almost two decades to accomplish, Linux did in half the time. It’s unheard of. Let’s come to terms with it. FreeBSD has flaws, issues and problems that need attention. And so does Linux. They both have their strengths and their weaknesses. And if we do know any of them, we shouldn’t be shy to expose them, just as the first poster has done. What particularly exasperates me is when people spew crap they have no knowledge of. If a particular implementation sucks, I wan to you tell me why it sucks and what should have been used instead and of course the advantages of your solution over what you think sucks. Telling me things like, “FreeBSD sucks”, or “Linux is not stable” or “udev shouldn’t be userspace” and the usual osnews comments. I’m happy FreeBSD exists. And I’d like to see more friendly and healthy competition between all the free and open implementations of Unix. When it comes to desktop utilities, I think Linux is ahead of most free and open implementations of Unix. At the kernel level, I really don’t know what all the fuss is about. In my mind, they are hardly miles apart from each other. Today, Linux has a little edge thanks to fate, and perhaps it’s license. It might continue on that path for a few more years. But nothing last forever. What I think might kill Linux or the *BSDs for that matter are the horrendous antiquated development tools. These tools are good and powerful but most of them are a pain in the but to use compared to those available on say, Windows or OS X. Except you find the open source *nix technologies fascinating, developing for them is not particularly attractive. 2003-12-25 6:06 am Two things give you away. First, your IP address being posted right there after your monikers, and second, your daft spelling of the word TURD. 2003-12-25 6:26 am How on earth could this comment even get left on here? It is plagued with misinformation and nonsense. “The O(n) ULE scheduler, which is supposedly better than the current scheduler FreeBSD is using, still isn’t finished. Meanwhile Linux has an O(1) scheduler.” I bet you don’t even know what O(n) or O(1) mean. Funny, because Jeff (originally) based ULE on Ingo Molnar’s O(1) scheduler. After talking to Jeff, Con Kolivas has said that ULE is going to be quite comparable with the linux scheduler. [ref: http://kerneltrap.org/node/view/1785 ] But I bet you already knew these things, right? “It also uses an inferior M:N threading model.” Obviously you have no background in computer science. Theoretically speaking M:N threading should be better than 1:1 — however it is harder to impliment. Once again I’m sure you knew this. “The Giant lock in FreeBSD, which keeps huge portions of the kernel locked to a single processor, is still used all over the place, [bla bla]” Right, Fine grain locking is a 5.3 priority. If you ever spent any time reading anything you’d already know this. “It has been left full of locking bugs [bla bla]” Irrelevent. You don’t know this, you’re just spreading FUD. Why haven’t the OSnews editors moderated this post down? This gives OSnews a bad reputation — personal opinion is one thing — but this guy is knowlingly spreading lies. 2003-12-25 6:48 am Every BSD article is starting to be the same. First poster tries to compare it to Linux and makes sure Linux looks better wether or not what he says is true. Then folks post things like well they both have their merits, which while true gets tiring to read (no offense guys, it’s just that trolls aren’t worth your time). Personally, I can’t wait for 5-stable. Once I can find a friend with high speed to download 5.2RC2 for me I can begin helping them test it out. Jared 2003-12-25 7:11 am Why haven’t the OSnews editors moderated this post down? This gives OSnews a bad reputation — personal opinion is one thing — but this guy is knowlingly spreading lies. Amen to that… what’s worse is he continues to spread the same lies from thread to thread, despite being told repeatedly that he’s wrong. I’m sure we’ll see more posts about the “O(n) ULE scheduler” from him in the future, as well as the “inferior M:N threading” of FreeBSD. 2003-12-25 7:42 am >Why haven’t the OSnews editors moderated this post down? This >gives OSnews a bad reputation — personal opinion is one >thing But the fact is that FreeBSD is really making slowwww progress. Linux on the other hand is scaling up fast very fast. Not only fast but also damn good. Linux is supported by the big guys, wich is important for the bussiness model so i will not use FreeBSD in my office. FreeBSD might become mainstream but Linux is already mainstream. I am not saying Linux is better or worse than FreeBSD but it used a lot more, is supported a lot more is commercialized a lot more.. People tend to choise, or nature, the second best products or protocols, like betamax vs vhs, so for LInux to succeed (20%/30% marketshare) it has to be not as prefect as FreeBSD 2003-12-25 8:00 am 5.2 is due out soon within a month and 5.3 development will begin afterrward. very soon! a very good news for me and other FreeBSD users, isn’t it? I can’t wait. RE:RE: Changes – Anonymous (IP: —.client.comcast.net) Well, actually Christmas isn’t for Christians. Therefore a very Christian in nature shouldn’t care about Christmas anyway. In fact, Christ was born mid-October, and he had been baptised on 6th or 7th January. The 25th Dec is actually sort of birthday of Sun god or something. (I dunno what it is called in English) Just enjoy the holidays. That is what I do as a non-Christian (no religion). 2003-12-25 8:02 am Scott Long is too harsh to his team when he says that releasing FreeBSD 5.3 later than June 2004 will discredit their project. I wonder where that fear comes from. Most people who read about this stuff know that the FreeBSD Release Engineering Team has its hands full with the new branch. Maybe Scott shouldn’t forget that there are no impatient shareholders eager to fire him in case he doesn’t deliver on schedule. 2003-12-25 10:15 am “What I think might kill Linux or the *BSDs for that matter are the horrendous antiquated development tools. These tools are good and powerful but most of them are a pain in the but to use compared to those available on say, Windows or OS X. Except you find the open source *nix technologies fascinating, developing for them is not particularly attractive.” I’ve used some of M$’s recent development tools and they produce horribly big executables – more than 100K for a very simple program. I’ve been told the reason GCC’s execs are smaller is because of shared libraries, well maybe MS should take a leaf out of their book? The idea if not the code itself? I suspect this is why those viruses which are floating around are so huge. There are IDEs for X by the way, like Anjuta. 2003-12-25 12:10 pm I too found that to be strange. I didn’t think the BSD people cared, its core users wouldn’t stop using it if FreeBSD 5 was pushed back another year. Has Linux 2.6 made them nervous? *shrugs* I’m pleasantly suprised that FreeBSD 5 is coming along at an increasing rate, anyone ever look into http://www.freebsd.org/smp sometime and see the list of remaining issues shrink over time? I keep noticing an Alan Cox being mentioned as the VM developer for FreeBSD 5, thats not the very same Alan Cox of Linux/Red Hat fame is it? I thought he was off at college now. 2003-12-25 12:49 pm Nice to see BSD coming along in a good way. One thing I’d really like though which is obviously overkill since BSD is mainly for servers…. but if it could configure a desktop environment (including autodetection for soundcard/gfx card) I’d be very pleased… I’m indeed lazy I know, but comfort is comfort is comfort. Oh, and I’m not interested if Linux can do this or that, I’m sure it can, but my life philosophy strongly disagrees with GPL, like it or not, but please respect it. Look forward to read more about your progress dear BSD 2003-12-25 12:51 pm First of all – I wonder when all Linux advocates and others will realize that FreeBSD philosophy does not contain word ‘market’. We are just making GOOD OS, pretty slowly, but more stable. For ‘but it used a lot more’ I can say only one thing – Windows is used MUCH more. So, Windows is better than Linux then? It is chiken and egg problem – FreeBSD developers don’t care of ‘market’, big companies need ‘market’ => FreeBSD does not have support from big companies. 2003-12-25 2:03 pm I got removable USB flash memory disk as an Xmas present. It says in the package that no driver is necessary in Windows ME/2000/XP, Mac OS 9.x, Linux kernel 2.4x operating system or higher. Doesn’t say anything about FreeBSD, though. I know that FreeBSD is targeted towards big server systems and that they don’t care an owl’s hoot about little people (single-pc home-user desktop-OS) like me, but the FreeBSD ports maintainers have done such tremendous job in porting desktop applications that FreeBSD has actually become my favourite desktop OS. I’ve noticed that some other crazy people also use FreeBSD as their desktop and from these people I’d like to ask if FreeBSD supports USB flash memory devices? If so, does it matter which version of FreeBSD you use? I’ve got 4.9 in my laptop – thought of making the big leap only when 5.3 comes out. Merry Xmas everybody & Hyvää joulua kaikille! 2003-12-25 3:20 pm By Matthew Smith (IP: —.in-addr.btopenworld.com) – Posted on “I’ve used some of M$’s recent development tools and they produce horribly big executables – more than 100K for a very simple program…. There are IDEs for X by the way, like Anjuta.” IDE tools like Delphi from Borland and Visual Stuido from MS are desperately needed for Linux. Meaning, tools that let you develop database front-end programs with BOUND DATA controls. A RAD system that’s fast to develop in. From what i see, Mono GTK# will deliver this. Just checked Anjuta and it doesn’t have any of this, so what’s its use to the corporate programmer? A couple of possible examples for linux now: one is PhpLens – it’s actually like Delphi, but for php – you can build very complex db front ends for a browser very very fast. impressive. Another example: Flash mx, since the plugin runs on linux, and does have bound data controls – it’s also auto-updateable on the client machine… 2003-12-25 4:39 pm OK, one more time, Keefer. Then please shut up – you’re telling lies, big lies, and I don’t understand why you deperately want to let FreeBSD look back. You’re a big troll. Stop spreading lies! In response to your comments: The O(n) ULE scheduler, which is supposedly better than the current scheduler FreeBSD is using, still isn’t finished. Meanwhile Linux has an O(1) scheduler. ULE is a O(1) scheduler, as I’ve told you before. It works and is very stable at this moment, and has excellent interactive performance. Something says to me that you haven’t actually tried it. 2003-12-25 4:58 pm KSE still isn’t finished and doesn’t work on Alpha and sparc64. It also uses an inferior M:N threading model. Meanwhile Linux’s 1:1 threading implementation, NPTL, is not only finished, but mature and has been in use for over a year, and works on Alpha and sparc64. If you look at the commit logs, you can see that there have been very little changes to KSE lately. Not because the team isn’t active, but because KSE is finished. All that is left is some bug hunting – and the FreeBSD team won’t make this the default until it’s absolutely stable. Alpha and SPARC64 support are a matter of time, seeing that IA64 and AMD64 already have support. Experts agree that M:N is, when well implemented, faster than 1:1. Look at the Tru64 threading libraries, for example. If you don’t like M:N threading or think that it is ‘inferior’ to 1:1, you can choose between 1:1 and M:N libraries (libthr and libkse respectively) on a per-application basis. It’s your choice, and you can do your own performance testing to get the best. If you have done some tests for your specific application that show that 1:1 threading work better for you, just use libthr for that application. The Giant lock in FreeBSD, which keeps huge portions of the kernel locked to a single processor, is still used all over the place, and huge portions of the kernel assume other portions are locked by Giant, leading to stability issues. It doesn’t look like you’ve actually used the FreeBSD 5.2 release candidates, because if you would have, you couldn’t have claimed that it is unstable. Things are being moved out from giant as we speak, and a large part of the drivers is already done (but you know that, right?). Performance on SMP systems is already getting very impressive, and the great thing is, it will only get better from now on as even more kernel drivers are moved away from the Giant Lock. The only way is up! Overall, it seems like FreeBSD 5.x is way too much new code. Linux got fine grained locking progressively betwen the 2.0, 2.2, 2.4, and 2.6 kernel releases, whereas FreeBSD 5.x tried to add it all in a single kernel version. It has been left full of locking bugs and other serious issues which should cause great concern to anyone considering using it over Linux for whatever reason. This is so ridiculous that I don’t think I have to comment . You’ve missed FreeBSD 5.0, 5.1 and 5.2, obviously. Also, the SMPng project has been working on this for years in -CURRENT, but seeing how little you know about these things, you probably don’t know what -CURRENT is either. Meanwhile, the Linux 2.6 kernel has already seen a stable release, and is scaling equal to or significantly better than FreeBSD in all respects, despite claims of the FreeBSD apologists that FreeBSD 5.x puts FreeBSD back on the same level as Linux. Linux 2.6 has completely blown away any basis for such claims. I’m awaiting your amazing benchmarks. Overall, it’s a shame that this troll (keefer) keeps posting this story on every FreeBSD post. It’s an even bigger shame that these lies (M:N being slower than 1:1, ULE being O(n), there not being a 1:1 threading library for FreeBSD – everyone can easily check that this is not true) are not moderated down on OSNews. I hope OSNews can stop these trolls, as the quality of this site (I’m talking about the comments, not the articles) seems to be decreasing lately. 2003-12-25 5:04 pm I’ve noticed that some other crazy people also use FreeBSD as their desktop and from these people I’d like to ask if FreeBSD supports USB flash memory devices? If so, does it matter which version of FreeBSD you use? I’ve got 4.9 in my laptop – thought of making the big leap only when 5.3 comes out. USB flash devices work by default, as they’re supported through the umass(4) driver. Just plug it in and you’re ready to go ! You’ll probably have to mount it as an msdos filesystem, as these things mostly come preformatted with FAT32. I don’t think it matters if you use 4.9 or 5.x. 2003-12-25 5:12 pm Nice to see BSD coming along in a good way. One thing I’d really like though which is obviously overkill since BSD is mainly for servers…. but if it could configure a desktop environment (including autodetection for soundcard/gfx card) I’d be very pleased… After you’ve done a standard install and started your FreeBSD system, type ‘kldload snd_driver’ as the root user to automatically recognize your sound card. If you want to make this permanent, add ‘snd_driver_load=”YES”‘ to /boot/loader.conf. You can also compile a custom kernel with ‘device pcm’ in your kernel configuration file, which will have the same effect (although it will take you longer ). The handbook has the details on this. Assuming you chose to install XFree86 when installing FreeBSD, you can use ‘XFree86 -configure’ to auto-generate a configuration file for X and recognize your video card. Follow the instructions to test if everything is OK. 2003-12-25 5:43 pm “FreeBSD has been stabler, faster, and generally more robust for a decade.” Really? My Debian and Slackware boxes have _never_ crashed in 6 years of use. I installed and fine-tuned FreeBSD but it offered no significant performance gain. Will you BSD fans please stop spreading FUD that FreeBSD is somehow “stabler and faster”? There’s absolutely no real-world proof of that, and you’re starting to look silly; Linux offers _much_ better performance on multi-processor machines and the 2.6 kernel is excellent on the desktop. Likewise with stability. Yeah, perhaps FreeBSD releases are more reliable than, say, Mandrake’s, but compare like with like — a community-run project such as Debian. There is no difference; they’re both extremely solid. Plus, Debian does much more extensive testing on its packages (compare to the bit-rot in FreeBSD’s Ports collection). If you feel the need to advocate FreeBSD, at least base it on real-world facts instead of vague and totally unproven allegations about stability and speed. Find something genuine to discuss, otherwise you’ll get shot down as follows: # Hardware support — Linux has a broader range of mature drivers, particularly on x86 # Ease of updating — Give me apt-get or yum over tiresome CVSuping and building and installing procedures any day # Software range — Most open source apps compile out-the-box on Linux, and there’s a growing amount of commercial software # Commercial support — MandrakeSoft supports its releases for 18 months. Debian for at least two years in practice. RHEL (and free variants like Tao Linux) offer a superb FIVE years of security fixes and patches. Meanwhile, FreeBSD has a measly 12 months. Yes, it’s not too hard to upgrade FreeBSD, but you end up with a slightly different system every 12 months; harder to target apps against Those are just a few points in which Linux soundly beats FreeBSD. And, coupled with the fact that a good Linux distro offers the same stability and performance of FreeBSD (if not better in some cases), there’s little incentive to run FreeBSD on x86 hardware today. 2003-12-25 5:46 pm P.S. I agree with the others. Why wasn;t this post moderated down? It is clearly nothing more than a troll who doesn’t know what he is talking about. It’s starting to seem that Slashdot has reached its carrying capacity, and a lot of the Slashdot fad wanker wannabes have migrated to OS News. 2003-12-25 5:51 pm “Will you BSD fans please stop spreading FUD that FreeBSD is somehow “stabler and faster”? There’s absolutely no real-world proof of that, and you’re starting to look silly;” There isn’t? Linux is the most popular Web server operating system in the world. Yet there is not one single Linux box on Netcraft’s longest uptime survery (average or maximum). What is on Netcraft’s longest uptime survey? Hmmm… What do you know. Every single site… All 50 of them, are running BSD. 2003-12-25 5:58 pm “# Hardware support — Linux has a broader range of mature drivers, particularly on x86” Would that be why FreeBSD supports more wireless networking devices than Linux does? Oh. And FreeBSD also had USB support long before Linux did. “# Ease of updating — Give me apt-get or yum over tiresome CVSuping and building and installing procedures any day” Give me “make world” any day. Can you rebuild your entire Linux system from source with a single command? “# Software range — Most open source apps compile out-the-box on Linux, and there’s a growing amount of commercial software ” Yeah. Once you figure out all the dependancies that you need to have installed, and then figure out how to get the linker to find those dependancies, etc. Sorry, But I think I will take FreeBSD’s ports “make install” any day for building apps from source. Grabs all the dependancies, recursively builds the dependancies from source, builts the app, and I am done. So if you want to trash FreeBSD, at least try to do it based on real world facts instead of the fact you don’t have a clue. 2003-12-25 6:06 pm “Here’s a little knowledge for you: Linux’s uptime counter wraps at 496 days. Now try to work out the rest.” Why? It sounds pretty stupid to me to wrap at 496 days. And why 496? What’s magic about that number? It’s not like it’s a variable size limitation or something. 2003-12-25 6:10 pm “And then ask yourself why Linux is so much more popular than the BSD flavours when it comes to web serving. Marketing?” Three reasons: 1. Commercial support, important to companies. 2. Marketing. FreeBSD generally doesn’t take out full page color adds in computer magazines like Red Hat does. 3. The fad popularity of Linux, which accounts for the fact that many of the Web servers that make Linux the most popular Web server platform are running in someone’s basement over a DSL line. BTW, when it comes to e-commerce, Linux is not the most popular platform. Windows actually is. 2003-12-25 6:23 pm “Why? It sounds pretty stupid to me to wrap at 496 days.” Hah, and you know what? It’s not half as stupid as you using Netcraft to try and show that FreeBSD is somehow more “stable”, and then making a complete mess of it. When advocating something, it’s best to know what you’re talking about first — otherwise you look foolish. “The fad popularity of Linux” Ah yes, of course. When someone runs Linux, it’s because they hate Windows and have been pulled in by all the advertising and don’t have a clue otherwise, right? And when someone runs BSD it’s always a purely technical decision? It’s easy to make such stereotypes. You know what? From my experience, people tend to run FreeBSD to be counter-culture; to be “different” and “better” than the Linux community because they use something which requires more time and effort. Many of us have real jobs to do. Going through the whole CVSup/build/install routine for each security issue, and having to update the OS every 12 months, is a waste of time and money. My Debian servers offer ultra simple one-command updating, and I know I can put them in place for at least two years without hassle. That, my friend, is why Linux is more popular. Above I listed some very crucial benefits that Linux offers over FreeBSD, and you haven’t been able to refute them (apart from some ill-informed nonsense about Netcraft). So, come on! Give us all some real-world, genuine and non-anecdotal advantages FreeBSD offers over Linux. I can name a few, such as a more tight developer community and very clean documentation, but they’re not as important as hardware support, multi-proc performance, software range and lengthy updates and fixes. And believe me, stability is incredibly important to me. But despite your unproven claims, good Linux distros offer just as good stability and performance as FreeBSD, with all the extra benefits on top. I’ve never had a kernel panic in 6 years. I know plenty of Linux users who can say the same. FreeBSD offers no reliability advantage, and is a lot weaker on many important fronts. Hey, it’s still a pretty good OS, but I haven’t found any concrete reasons to run it over Linux. 2003-12-25 6:33 pm “Would that be why FreeBSD supports more wireless networking devices than Linux does? Oh. And FreeBSD also had USB support long before Linux did.” Did I say Linux supported _everything_ better than FreeBSD? No. So feel fine to pluck out very specific examples, but on x86 hardware Linux has a much broader and more mature range of drivers. Live with it. As for the USB comment, that’s laughable. So what? Windows had USB support before both of them — does that somehow make it better then? Oh dear! “Give me “make world” any day. Can you rebuild your entire Linux system from source with a single command?” Nope, and I can’t repaint my house in the dark whilst juggling on a unicycle. Hint: it’s not something anybody needs to do on a regular basis (developers aside). When I’m deploying servers and workstations, being able to perform tricks like that is redundant and entirely unimportant. It’s a neat feature, yes, but the FreeBSD developers should concentrate on things that matter in the real world… (Also note that needing a development toolchain on a production server is very poor. FreeBSD really, really needs binary updates pronto.) “Yeah. Once you figure out all the dependancies that you need to have installed, and then figure out how to get the linker to find those dependancies, etc.” Sigh. Go read up on Apt-FU and Gentoo’s Portage system. They can do everything Ports can, and sometimes more. As said before, Ports is pretty good but there’s a huge amount of bitrot in there and packages marked as broken. 2003-12-25 6:44 pm “Ah yes, of course. When someone runs Linux, it’s because they hate Windows and have been pulled in by all the advertising and don’t have a clue otherwise, right?” Not all the time. But given the comments by many of the people on OSNews, I think it is fair to say that yet, a lot of people run Linux because of a “screw the establishment” attitude. “From my experience, people tend to run FreeBSD to be counter-culture; to be “different” and “better” than the Linux community because they use something which requires more time and effort.” I also see in FreeBSD, a development process that is more professional, and a core team that is more likely to know what they were doing. I’m sorry if I don’t trust an 18 year old kernel maintainer who doesn’t even have a BS degree in computer science when it comes to my mission critical applications. (Ok. So he is 20 now). “Going through the whole CVSup/build/install routine for each security issue, and having to update the OS every 12 months, is a waste of time and money.” The process is so incredibly simple that you can even automate it if you want to. That might not be the smartest thing to do, but the point is you can. So I don’t see what you are complaining about? And I would rather update the entire OS then bits and pieces like you tend to do in Linux. It’s far easier to maintain a standard FreeBSD installation across a few thousand servers than a standard Linux installation. That’s a complaint I often here from Linux admins. They have 200 servers, and probably none of them have the exact same OS installed because of partial updating and such. That, to me, makes FreeBSD less maintaince intensive on large server farms than Linux is. Because I always know the exact OS version of my FreeBSD systems. I don’t know that for my Linux systems because too often, some systems get one or another patch installed, and others don’t, etc. And I hear that as a common complaint from Linux admins on large server farms. 2003-12-25 6:48 pm “Did I say Linux supported _everything_ better than FreeBSD? No. So feel fine to pluck out very specific examples, but on x86 hardware Linux has a much broader and more mature range of drivers. Live with it.” True. But for severs, FreeBSD supports more of the hardware that is important. Sure, FreeBSD might not support the latest 3D graphics accelarator, and Linux might support it. But on servers, who cares? I pluck out specific examples that are important to the role in question. Wireless network support is important for servers. And FreeBSD supports it better than Linux does. I would also argue that FreeBSD’s IPFW is far easier to configure than Linux’s firewall, and that VPN in FreeBSD is far easier to configure than in Linux. And one last real advantage FreeBSD still has over Linux. The Berkely TCP/IP stack is still the best TCP/IP stack ever designed, which is why many many commercial operating systems choose to use it. Linux chose not to, and the Linux TCP/IP stack is not as strong as the Berkely TCP/IP stack. That’s also important in networking. 2003-12-25 6:51 pm “When I’m deploying servers and workstations, being able to perform tricks like that is redundant and entirely unimportant. It’s a neat feature, yes, but the FreeBSD developers should concentrate on things that matter in the real world… ” I do think it matters in the real world when you are dealing with a large server farm. Because it is an easy way to ensure that all of your servers are configured with the same core OS. That’s not as easy to do in Linux, as I pointed out earlier. And once again, I don’t see what the big deal is. Updating FreeBSD is not that complicated. It’s so simple it can be completely automated. 2003-12-25 6:54 pm “As said before, Ports is pretty good but there’s a huge amount of bitrot in there and packages marked as broken.” But almost all of the broken packages are desktop or workstation packages. Not packages that would be running on the typical network server. I agree that Linux is probably a better choice for a workstation or desktop because yes, many of the ports of desktop apps and such in FreeBSD are often broken. But almost all of the packages that are important to network servers work. And when they are broken, they usually get fixed very quickly. 2003-12-25 6:57 pm “Not all the time. But given the comments by many of the people on OSNews, I think it is fair to say that yet, a lot of people run Linux because of a “screw the establishment” attitude.” They do. And from comments on Slashdot and freebsd-advocacy it’s clear that a lot of people run FreeBSD to be “counter-culture” and “l33t3r” than Linux users. So let’s avoid making judgements based on hyperactive kids on message boards! “I also see in FreeBSD, a development process that is more professional, and a core team that is more likely to know what they were doing.” Define “professional”. Microsoft employs more professionals than there are FreeBSD and Linux kernel hackers put together, but the end result isn’t necessarily better. Your “professional” core team just kicked out one of your most talented developers, Matt Dillon, all the time muttering away on their closed personal list and leaving end-users in the dark. So it’s clear that “professional” doesn’t always lead to good code. There are massive flamewars and childish insults on linux-kernel, but at the end of the day they care about _good_code_. And that’s what I care about, and I don’t want to see great developers leave because of petty infighting. That has been a long-running problem in the BSD communities, and Matt’s new project has diluted the development effort even further. “The process is so incredibly simple that you can even automate it if you want to. That might not be the smartest thing to do, but the point is you can. So I don’t see what you are complaining about?” But it’s _still_ more hassle than a single “apt-get” command, isn’t it? It takes more time, and time is money in business. Plus, it means you need the development toolchain on a box which adds security problems. “They have 200 servers, and probably none of them have the exact same OS installed because of partial updating and such.” That’s not remotely a fault in Linux distributions though; if an admin picks-and-chooses his patches, he’s bound to end up with different systems. Just as a FreeBSD user could selectively apply patches. Not sure what you’re getting at here. If I install 20 Debian servers, and set them all apt-get updating, they’ll all be exactly the same once the process has completed. “That, to me, makes FreeBSD less maintaince intensive on large server farms than Linux is.” To me, maintainance means I don’t have to upgrade a bunch of machines every year — as said, Debian supports their distros for at least two years, and RHEL offer a whopping five. Understandably, it’s harder for the FreeBSD developers to do that as they have limited resources, but it’s a crucial area in which Linux is very far ahead. 2003-12-25 7:01 pm “But almost all of the broken packages are desktop or workstation packages. Not packages that would be running on the typical network server.” Righto. So this adds an interesting slant on the whole “stability” thing too — if we’re talking about the OS as a whole and not just the kernel/C library etc., a FreeBSD is likely to be _less_ reliable than a good Debian installation simply because the latter has had much longer testing and patching. As you’ve agreed, there are unmaintained and broken packages. That’s not good for all-round stability. 2003-12-25 7:16 pm I suggest some of you stop the trolling, baiting and angry tone around here, or you will see the whole thread deleted. 2003-12-25 7:24 pm As a FreeBSD person, I agree with you: we should all stop spreading FUD. Except that you haven’t done quite enough research (which isn’t at all your fault, since you’re a Linux user and probably don’t know about this). – FreeBSD has a binary updater for the main system and kernel (‘world’). It is maintained by Colin Percival and can be installed from the security/freebsd-update. You don’t need a development tree for this. – I don’t agree about the ‘many broken packages’. Packages that are broken are marked broken. EOT. – If you don’t update your system for two years, you’re going to leave security holes. – The ports/packages system is fantastic when you’re administering multiple servers; build the ports on one machine, create packages with pkg_create, use them to set up your servers, and you’ll have the exact same package on all machines. This way you keep complete control, you can apply custom patches or options, and updating is extremely easy. 2003-12-25 7:27 pm The 497 day thing happens in most current OSes (FreeBSD 4+ included) and is obviously related to using a 32 bit number to store the time (just check the ms). However, lots of other info from netcraft praises FreeBSD over any other OS – check the availabilty section. Look, there just isn’t anything that would make someone who’s fond of neither prefer Linux over FreeBSD as an OS: FreeBSD is tidier, a lot easier to install and configure and maintain, faster (you’ll realise it once you get to use it to do actual work instead of devising dumb benchmarks) and overall a lot more enjoyable. FreeBSD has exactly two defects: . Unability to install to a logical partition . Irritating habit of setting ts slice as the active partition when we tell it not to touch the MBR (if we tell so, it’s because we want to leave the boot process as is, nu?) . Less guaranteed hardware support (which Linux is no wonder either compared to Windows) Linux in many ways parallels Windows in its design choices: go the easiest way ahead and then someday hack it so it doesn’t explode (VM, scheduler, monolothic…) Fact is Linux is hype. And it’s good that it is, because that way there’s lots more free software out there. If it weren’t for Linux, we would never have heard of the FSF and there would be no such thing as the Ports system. 2003-12-25 7:27 pm Considering that the first post by Keefer is the biggest troll in this thread, you might want to start the deletion process with that post. Adam 2003-12-25 7:40 pm Just for the record you can install/update packages on FreeBSD with one simple command. You just do #pkg_add -r ftp://favoritemirror/path/package.tbz and it installs the package and its dependancies from the ftp server (likewise for pkg_update). Just like your fancy debian, now isn’t that nice 🙂 2003-12-25 7:46 pm As of late I’ve seen a lot of harmless messages modded (just because they had something OT or used any risible word found offensive) while obvious candidates like people who insist on blatant technical lies hinting even that they know they’re FUDding pass the test. I attribute the latter to an admin’s fear of being called a ‘censor’. So, harmless things that may displease the troll are modded, while harmful intentional arrogant FUD is left untouched lest someone says there is no freedom of speech. 2003-12-25 7:46 pm “FreeBSD has a binary updater for the main system and kernel (‘world’). It is maintained by Colin Percival” Yep, but no business is going to trust it. That needs to be an integral part of FreeBSD first; large companies aren’t going to rely on something developed by a single guy in his spare time. “If you don’t update your system for two years, you’re going to leave security holes” With FreeBSD, you don’t have the option — you have to update every 12 months, and re-test all the apps you’ve built around it (as components like Perl etc. change underneath). With Linux, you have a choice — up to 5 years if necessary. This is very important when you have a lot of servers to admin. “FreeBSD is tidier” Try Slackware. And NetBSD is “tidier” than FreeBSD; your point? “a lot easier to install and configure” Erm, how? I prefer the command line, but take something like Red Hat’s tools — orders of magnitude easier than Sysinstall and its handful of sub-tools. “and maintain” If you really think FreeBSD’s update system is somehow easier/quicker/simpler than apt-get, you must be on a different planet to me! “faster (you’ll realise it once you get to use it to do actual work instead of devising dumb benchmarks)” Oh dear. Your patronising attitude aside, how can you make such claims without benchmarks? Just anecdotal “evidence”? It’s funny you should say “actual work”. You see, I’ve worked with a few 4-CPU servers and Linux totally obliterates FreeBSD on the performance front. And multi-proc boxes are becoming increasingly common. Who’s doing “actual work” now, eh? “Fact is Linux is hype” Fact is, Linux is going everywhere. Google don’t run it on 10,000 boxes because of “hype”. It’s very stable, faster than FreeBSD on a lot of hardware, has longer and better supported releases, works with a wider variety of hardware, runs more software and is easier to use. Nobody here has been able to refute those facts. Just keep making completely vague claims like “FreeBSD is faster” and “Linux is hype”, when in the real world people will continue to deploy Linux and enjoy its excellent support, wide software availability, and friendly communities. 2003-12-25 8:10 pm >>FreeBSD is tidier > >Try Slackware. And NetBSD is “tidier” than FreeBSD; your point? “Tidier” => easier to configure and maintain. I’m not comparing FreeBSD to NetBSD. >>a lot easier to install and configure” > >Erm, how? I prefer the command line, but take something like Red Hat’s tools — orders of magnitude easier than Sysinstall and its handful of sub-tools. I’ll grant that sysinstall isn’t marvelous. However, prefering the CL as well, I find it easier to configure FreeBSD than any Linux. >>and maintain > >If you really think FreeBSD’s update system is somehow easier/quicker/simpler than apt-get, you must be on a different planet to me! I like apt-get. I like Debian and generally anyhting Debian-based a lot. But just as configure/make/make install beats any install shield wizard, I find FreeBSD’s update mechanisms more palatable. >>faster (you’ll realise it once you get to use it to do actual work instead of devising dumb benchmarks) > >Oh dear. Your patronising attitude aside, how can you make such claims without benchmarks? Just anecdotal “evidence”? Patronising (who chose the word anecdotal)? Every freshman CS student should know what benchmarks are worth and that’s exactly what I mean: there is no other way to test something rather than actually doing it. If you want to measure a server’s performance, you must do it by running a site, not an accept()/fork() benchmark. So, since you say that on 4-CPU servers FreeBSD is obliterated by Linux, I’ll have to believe it or do it myself. >>Fact is Linux is hype” > >Fact is, Linux is going everywhere. Google don’t run it on 10,000 boxes because of “hype”. No, they run it because it works. Hype is orthogonal to that. >It’s very stable, faster than FreeBSD on a lot of hardware, has longer and better supported releases, works with a wider variety of hardware, runs more software and is easier to use. Aside from 1, 3, and half of 4 (I generally find FreeBSD’s support for communications hardware to be more reliable), 5 is wrong and 6 is too vague (what use? non-OS userland is the same; config has been discussed above; FreeBSD OS userland generally has more functionality). >Nobody here has been able to refute those facts. Just keep making completely vague claims like “FreeBSD is faster” and “Linux is hype”, when in the real world people will continue to deploy Linux and enjoy its excellent support, wide software availability, and friendly communities. In the real world, Windows and Linux will keep eating the market share of commercial UNIX while the BSDs will remain where they are (unless Apple suddenly starts selling more and more laptops and Sun gets poorer countries to buy their desktop). 2003-12-25 8:52 pm “I’ll grant that sysinstall isn’t marvelous. However, prefering the CL as well, I find it easier to configure FreeBSD than any Linux.” I’d agree if we were talking about Red Hat and its convoluted init scripts etc., but saying _any_ Linux is a bit OTT. Slackware is just as clean and simple as FreeBSD under the hood. “So, since you say that on 4-CPU servers FreeBSD is obliterated by Linux, I’ll have to believe it or do it myself. ” It’s not just me. Key FreeBSD engineers in the SMPng effort have said that they’re only just up to Linux 2.2 in performance. “In the real world, Windows and Linux will keep eating the market share of commercial UNIX while the BSDs will remain where they are” So ask yourself _why_. Sure, there’s a lot of hype around Linux at the moment, but that wasn’t always the case. Why, early on, did Linux gather attention far quicker than FreeBSD? Many would agree that the Linux development process was more open and approachable than the elite BSD club. Licensing played a large part. But to dismiss recent massive Linux deployments as “hype” is simply naive and offensive to those who’ve thought it through. Yes, maybe FreeBSD should be considered more often, but you can’t blame a company/govt for choosing the OS which has better and longer support schemes, broader software availability and is easier to use (on the desktop definitely, with modern distro tools). As I keep saying, I’ve nothing against FreeBSD; it’s a great OS. But things like support, drivers, software ability and easy upgrading matter, and Linux has the edge in these areas. And, set up right, it’s just as reliable as FreeBSD (my Linux servers and workstations NEVER crash) so that FUD needs to be stopped. 2003-12-25 9:27 pm > I’d agree if we were talking about Red Hat and its > convoluted init scripts etc., but saying _any_ Linux > is a bit OTT. Slackware is just as clean and simple > as FreeBSD under the hood. Slackware config is usually praised and why? because it’s clean and simple as you point out. Oddly, Slackware is considered the least user friendly of Linuxes. What I like in FreeBSD is that the idea of tidiness pervades the whole system. In Linux at the very kernel there’s clean parts, well laid desins, and there’s much that isn’t so. > But to dismiss recent massive Linux deployments as “hype” > is simply naive and offensive to those who’ve thought > it through. I did say Linux is hype – that doesn’t amount to dismissing it. While I think that, for instance, Java is all hype and little substance (and even so, it’s a step forward from C++), I’ll agree that Linux has quite a lot of substance to it, and as I proceeded to say, the hype is a good thing – without Linux fewer would have heard about the BSDs and there wouldn’t be anywhere as much free software. I don’t know any BSD user with angry feelings towards Linux, or considering it harmful. I’ve seen the reverse. I also regret the quarrels within BSD development. Some people appear to have way too many attitude problems. But such quarrels exist everywhere where forking is unconceiveable as well. 2003-12-25 9:58 pm large companies aren’t going to rely on something developed by a single guy in his spare time. replace single guy with “bunch of guys” and you’ve summed up debian. I’ve nothing against FreeBSD you had me fooled Has anyone here ever read Ken Thompson’s comments on Linux? Computer: In a sense, Linux is following in this tradition. Any thoughts on this phenomenon? (from ’99 so a lot of it is admittedly out of date) Thompson: I view Linux as something that’s not Microsoft—a backlash against Microsoft, no more and no less. I don’t think it will be very successful in the long run. I’ve looked at the source and there are pieces that are good and pieces that are not. A whole bunch of random people have contributed to this source, and the quality varies drastically. My experience and some of my friends’ experience is that Linux is quite unreliable. Microsoft is really unreliable but Linux is worse. In a non-PC environment, it just won’t hold up. If you’re using it on a single box, that’s one thing. But if you want to use Linux in firewalls, gateways, embedded systems, and so on, it has a long way to go. Just remember that were it not for BSD the internet would not be what it is today. 2003-12-25 10:35 pm “or you will see the whole thread deleted” I personally think that that is not a bad idea. Too many trolls in this thread. Delete it. 2003-12-25 11:46 pm “replace single guy with “bunch of guys” and you’ve summed up debian” And? Businesses don’t trust Debian as much as, say, Red Hat and RHEL. What’s your point? I’m saying that while cpercival’s system is good, it needs to be part of FreeBSD as a whole before it’s to be relied upon. “you had me fooled ” That’s because whenever anyone says something less than positive about FreeBSD, the BSD fans take it all so personal and get irate. I’m not sure why, but you certainly feel threatened by Linux. Again, note that I said FreeBSD is a good OS but it’s not strong in certain areas, and now you take me to be some sort of BSD-hater. I’ve probably run more BSD systems than you’ve had hot dinners; I’ve made modifications to the FreeBSD kernel and done loads with the OS. “Has anyone here ever read Ken Thompson’s comments on Linux? ” Has anyone read Bill Gates’ comments on Linux? What does it matter how a single person thinks? Technically knowledgable people the world over are using and deploying Linux because it’s fast, stable and well-supported; don’t dismiss them. “Just remember that were it not for BSD the internet would not be what it is today.” Remember without William Shockley computers would not be what they are today. Doesn’t mean he’s a major relevant force in computing TODAY though… As for what Kingston said… “I personally think that that is not a bad idea. Too many trolls in this thread. Delete it.” If you think someone trying to establish the truth, rather than letting the FreeBSD fanboys delude themselves with ficticious talk of Linux’s “instability” and “slow performance”, is trolling, then you have a different definition. I never said “BSD is dying” or anything like that. In fact, I praised many aspects of the OS. That’s not trolling, and Eugenia is usually open to technical debate on this site so it’d be sad if she deleted this thread. I’m just defending Linux from FUD and inaccuracies. Remember, BSD fans, I was once like you — I ran FreeBSD everywhere, thinking it was cool and different to not be using the increasingly-popular Linux. I had the placebo effect of listening to other BSD users, saying that it was “more stable” and “faster” than Linux, and just lapped it up. Then, one day, I decided to see how the Linux world was progressing. So I installed Slackware. And I found a system that was just as clean, rock-solid and fast as FreeBSD, but with other benefits on top. I thought, “What exactly is FreeBSD offering me? It’s a good system but there are few significant reasons to use it over Linux”. And it was a mini revelation. Choosing the tool that helped me work better, not what made me feel l33t, counter-culture or different. I had the stability and performance of FreeBSD, together with better support, broader hardware compatibility and a wider software range. And all with the reliability and performance I’d enjoyed from FreeBSD. So let’s just stop the FUD about Linux; that’s all I’m trying to do. Yes, FreeBSD is a very good OS with some great features, but a mature Linux distro offers just about all of them and extra benefits. The two OSes can happily go on, but advocate the real benefits of FreeBSD and not vague, unproven anecdotes. Because you’ll only turn away potential converts, if they find that their system _isn’t_ “more stable” or “faster”. 2003-12-26 12:30 am If I install 20 Debian servers, and set them all apt-get updating, they’ll all be exactly the same once the process has completed. Not if the repository changes it’s contents after the first 10 servers. In any case, I gave up on Debian the first time I tried to install ssh and ended up with a list of 10 other packages that had a circular dependency. Unfortunately, one of these 10 wasn’t even available on the server, making it impossible to install ssh. For nearly two weeks this continued. Needless to say, I gave up on Debian. Adam 2003-12-26 1:10 am a wider software range What is the reason you say this? besides java(it is progressing),oracle(tho you could run under linux binary emulation. Other than on Red Hat and SUSE linux, oracle is not supported by the vendor anyways) and other few uncommon linux only applications what is short on FreeBSD in the term of software range? Could you please name a few? I am just asking. And Slackware is based on BSD system; its init/rc system and it is why the slackware is just as clean as FreeBSD. 2003-12-26 1:36 am Stop impersonating me. Retarded numskull. Eugenia, can we please get username-registration or something? This is simply annoying. 2003-12-26 2:19 am I feel like I should post a comment. After all, every stupid dipshit could do that too. This one goes to Realist. Yeah, I’m sure you operate a humongus server farm. Oh, and you sure did some improvements to FreeBSD kernel (care to tell us if you submitted them back to the dev team ? If so, which ones so we can all see and belive, after all anybody could just say so). Yes, FreeBSD sucks because of Debian is so grat with apt system. Nevermind that you can use ‘pkg_add -r’, or portupgrade. You wanna get real, mr. Realist ? Nobody in a corporte world gives shit about Debian. If they run GNU/Linux on their hardware, it’s RedHat or SuSE. And quite frankly RPMS suck. You wanna give some real world examples of why GNU/Linux is better than FreeBSD ? Because I’m sick of your pulled out of your ass examples like ‘oh, yeah, linux is great on that 4 proc system’ or ‘it scales so good compared to FreeBSD’. Right. Notice how I keep using GNU/Linux. That’s right, Linux is just a kernel. You still need all the apps to go with it buddy. FreeBSD provides everything you need for a complete Unix experience. 2003-12-26 2:59 am I’ve probably run more BSD systems than you’ve had hot dinners; yeah right, you’ve never even met me. I’ve made modifications to the FreeBSD kernel and done loads with the OS. care to elaborate? I’m not sure why, but you certainly feel threatened by Linux. nope sorry I don’t think about it much anymore since I’ve stopped using it entirely. Remember I’m not on a Linux thread listing reasons as to why FreeBSD is better. all in all I really don’t care if you want to run whatever Linux distro suits your fancy. but I just had to respond to that hot dinners comment. Jared 2003-12-26 4:02 am I appreciate, the comment I really liked it. 2003-12-26 5:40 am Linux 2.6, 2.4, FreeBSD -CURRENT, NetBSD, OpenBSD http://bulk.fefe.de/scalability/ Essentially it shows that linux 2.6 (in these tests) is much better than any of the BSDs, and overall it looks like (best to worst) L2.6, L2.4/FBSD-C,(very close, I would say equal) NBSD, OBSD. Initially now it looks like L2.6, NBSD-C, FBSD-C/L2.4, NBSD, OBSD-C, OBSD. Please note that the ordering is My opinion, and the site doesn’t really comment on it more than 2.6 is the clear winner in most tests. Frankly, I would like to run gentoo with all of them (preferably on the same fs etc and test them) unfortunately most of the BSDs don’t support many of the filesystems linux does, so this may be a while, but I hope it can be done, and select FBSD kernel, OBSD kernel, NBSD kernel, Linux 2.6 kernel & Linux 2.4 kernel and compare them fairly on the exact same system (it likely won’t happen, but it would be nice, and seriously reduce fragmentation.) 2003-12-26 6:58 am > Essentially it shows that linux 2.6 (in these tests) > is much better than any of the BSDs No, what it does ‘show’ and conclude is that L2.6 is overall the best but FreeBSD 5.x is very close and on a par with it, while L2.4 is way behind and both NetBSD and OpenBSD are terrible. > http://bulk.fefe.de/scalability/ Besides, there’s something broken with at least some of these benchmarks. I’ve looked at them when they originally were posted (a week or so) and if you take the time to read the addenda more carefully you’ll see the author was (unintentionally) unfair to the BSDs, and apparently believes that the NetBSD team managed to improve the performance in one of them from O(n2) to O(1) in one week. This and other small things (which the author admits) clearly suggest that either there’s some wrong with the tests or, as I said above, benchmarks in general are worth very little. > most of the BSDs don’t support many of the filesystems > linux does What are you thinking of? Well, I don’t know how many fs ekkoBSD supports… but once again, any BSD is an OS, not a kernel (Debian/BSD projects notwithstanding). 2003-12-26 7:42 am Essentially it shows that linux 2.6 (in these tests) is much better than any of the BSDs, and overall it looks like (best to worst) L2.6, L2.4/FBSD-C,(very close, I would say equal) NBSD, OBSD. Initially now it looks like L2.6, NBSD-C, FBSD-C/L2.4, NBSD, OBSD-C, OBSD. That’s a highly flawed interpretation of the results, and furthermore these tests are quite dubious to begin with. First, the ULE scheduler in FreeBSD was not enabled. Secondly, FreeBSD beat Linux 2.4 in virtually all of the tests: socket() performace (which shows better overall file descriptor table management), fork() performance, and access latency of mmaped pages. The only tests in which Linux 2.4 came out ahead are the “HTTP access latency” and the initial mmap() speed tests. What daemons are you running that use a signal-driven I/O model. Unless you answer something like “Hoser FTPD”, the answer is probably none. What daemons are you running that are making mmap() calls more often than they’re accessing mmap()ed pages? In the case of FreeBSD versus Linux 2.4, it appears that FreeBSD has traded more overhead in the initial mmap() call for faster page access speeds… faster than Linux 2.6. However, Linux 2.6 has found a way to completely mitigate the O(n) nature of the initial mmap() call exhibited by both FreeBSD and Linux 2.4, but still loses to FreeBSD in page access latencies. Furthermore, in order to provide O(1) I/O multiplexing on par with FreeBSD’s kqueue() mechanism, signal-driven I/O must be used. This involves asynchronous event handling, which is a particularly messy process. Consequently, this requires applications custom tailored to an asynchoronus event model, as opposed to kqueue() which can serve as an effective “drop-in” replacement for O(n) mechanisms such as select() or poll(). Due to its long-time availability, kqueues are already supported by a number of transactional servers, most notably ircd, while support for Linux’s epoll is forthcoming in most ircd implementations. Supporting sigio on Linux 2.4 would almost certainly entail a complete rewrite. Due to the need to use signal-driven I/O in order to scale O(1) in multiplexing operations, still scaling O(n) in mmap() calls, and slower access times for mmap()ed pages, Linux 2.4 is clearly the loser when compared to FreeBSD 5.x. The only place where Linux 2.6 is a clear winner is scaling O(1) in initial mmap() calls, and even that is a minor victory due to the slower page access latencies. In conjunction with the ULE scheduler and KSE threading, FreeBSD’s scalability on x86 and AMD64 servers is on par with Linux 2.6 in virtually all respects. 2003-12-26 10:18 am Fiddly: that is what FreeBSD is. The docs are good, but count for nothing when I have to hotwire GDM and rc.conf or rc.local or whatnot. From a server perspective, Fedora vs FreeBSD. FreeBSD is … again very fiddly. I’ve set up quite a number of servers for my company, and this was when I was migrating from Windows to *nix. FreeBSD was a huge mistake. Took me soooooo damn long to get it up and running. Debian was like 30 minutes (inclusive of setting up Samba shares/users etc). My new choice of server is however Fedora. On the desktop and on the server, Linux 2.6 is going to take the world by storm. No question about it. And the BSD users that are so damn immature and unfriendly (FreeBSD – OpenBSD and NetBSD were friendly, and so was Debian) are just going to bitrot as they lose more and more installations. I really don’t care, but when there is a flame between FreeBSD and Linux, Linux tends to make sense, and BSD seems to be all defensive – Guiltily so. Personally, I don’t care what kernel is running under, just want I can get done; Linux Gets Shit Done. 2003-12-26 10:41 am Judging from this thread… I’d say those with some experience in Unix tend to go for BSD and those new to Unix tend to choose Linux and try distros until they get mad. Still, almost all BSD users understand the value of diversity. Linux advocates are the reason why I’m not interested, and yes you’re ‘causing hype. 95% of the servers out there are not in need of scaling 32 CPU’s…. they hardly need 2. What they need is RELIABILITY and low administration. Reinstalling distros and configuring yourself to death is not cost-efficient. This is the answer to why BSD exists! 2003-12-26 11:44 am > I really don’t care, but when there is a flame between FreeBSD and Linux, Linux tends to make sense, and BSD seems to be all defensive – Guiltily so. No, no, and again no. BSD seems to try to explain those linux zealots that they need to read and understand BSD. They know nothing about scheduler, new developments in 5.x branch, but tend to speak (and not only tend :>) about it. BSD position is ‘we are trying to make you know things you are talking about’. I personally use both Linux and FreeBSD here. Debian and FreeBSD, and some small count of RedHat – it is being replaced with Debian on ocasion. I know both of them very well, and I really don’t care what to run – I need things to be done. If I see that task could be done better with Linux – I’m using linux, if FreeBSD will serve it better – I will use FreeBSD. The worse thing is that Linux is hype, too much idiots, which are not ‘for Linux’, but ‘against Microsoft’ or ‘against *BSD’. FreeBSD has completely different approach – we are not ‘for’ or ‘against’ something, we are just making our OS better. In any possible way – some write documentation, some – code, some – problem reports. All FreeBSD users (not neccessarily developers) around the globe are one single strong TEAM. 2003-12-26 1:12 pm You are off topic. Personally, I don’t care what kernel is running under, just want I can get done; Linux Gets Shit Done. What things can’t you get done on FreeBSD while you can get them done on Linux? Could you name them? because.. It could be only your problem! Others may get them done better on FreeBSD. For me I can get most of my tasks done better and faster on FreeBSD than on Linux. Don’t make FUD :p And Fedora is not intented for server system. How on earth could your new choice of server be Fedora? Whether or not its a new choice, I doubt you could use it as server. Well, for the personal server that maybe fine but for company.. ummm nope! (fast changing unstable desktop OS on critical server system is not a company wants) FreeBSD was a huge mistake. Took me soooooo damn long to get it up and running. Debian was like 30 minutes (inclusive of setting up Samba shares/users etc) What is the huge mistake of FreeBSD? What I can see is your mistake. It doesn’t take me soooooo damn long to get FreeBSD up and running. But wait a seconds. How long is actually ‘soooooo damn long’? Have you ever looked up the Handbook before starting installation? FreeBSD is FreeBSD not Linux. Just with your Linux (probably only Debian and RHL specific) knowledge you cannnot success on FreeBSD. But don’t blame FreeBSD just because you fail to get it up and running. It’s again your own problem. Blame yourself and grow up. This article was about the new roadmap for 5.3 is going to come out soon. Therefore the fact is you trolling with FUD. 2003-12-26 1:46 pm “What things can’t you get done on FreeBSD while you can get them done on Linux?” 1) Installing a server and leaving it in place for two to five years, applying patches remotely. Impossible with FreeBSD; you need a whole system update after 12 months 2) Running critical applications at acceptable performance on multi-processor hardware 3) Various specific tasks demanding certain drivers for which FreeBSD’s support is weak or non-existant 4) Running various commercial applications natively, and supported, instead of risking them _unsupported_ via an emulation layer 5) Be guaranteed that software development decisions are made on technical grounds in an _open_ forum (as opposed to members of a closed list bickering and throwing out a very talented coder, diluting their efforts) 6) Introduce inexperienced users to the benefits of free/open source software. What happened to libh, FreeBSDers? See, there’s loads of stuff Linux can do that FreeBSD can’t, much of it vitally important to business (and the reason why Linux has seen much more take-up; it’s not down to “hype”). What can FreeBSD do? Ooooooh, “make world”? Businesses, home users and non-developers don’t need that! “Better TCP/IP stack”? Subjective, and the differences are negligible at most! So, what can it do? If it supports less mainstream hardware that Linux, runs fewer commercial apps in a supported configuration, has updates for a much shorter space of time (12 months? ouch), and is no more stable than a mature Linux distro, what can it offer? 2003-12-26 1:51 pm > Running various commercial applications natively, > and supported, instead of risking them _unsupported_ > via an emulation layer Even you should know better than that; it’s not usual emulation, it’s API-mapping. I run Win32 software under Odin better than it runs under Windows. I’ve ran Linux software under FreeBSD better than under Linux… 2003-12-26 2:58 pm “Even you should know better than that; it’s not usual emulation, it’s API-mapping.” Yep, I know that, but for the sake of discussion I wasn’t going to be pedantic about it. And yes, I’ve had plenty of success with FreeBSD’s Linux compatibility subsystem; however, that’s fine for small scale use but the original poster asked for things Linux could do which FreeBSD couldn’t. In most cases you will not get support for a business application running under an emulation layer (or, at least, very much reduced support). So for running certified, critical commercial applications natively and supported, Linux is a better choice. (And before someone points out that softcos support native Windows apps even more, that wasn’t the comparison. This is between FreeBSD and Linux.) I’d also like to point out, for those getting their knickers in a twist and labelling me as some BSD-hater, I’m a big NetBSD fan and have also put OpenBSD to good use. Both systems offer some significant advantages over Linux and, pleasingly, advocates of both OSes realise that they’re not appropriate for all tasks. Meanwhile, FreeBSD, in attempting to be an general purpose x86-centric OS, is striving (mostly) for the same goals as Linux. NetBSD offers superb portability – pretty much unmatched in the free OS world. OpenBSD works incredibly hard on security – again, surpassing other free OSes. But FreeBSD aims for performance and stability on x86, a valid goal, but Linux already offers that with the other benefits I’ve mentioned (hardware support, longer updates etc.). So when FreeBSD fans spread FUD and say that Linux users are just pulled in by “hype”, it’s offensive; for the vast majority of cases, Linux is the better choice. Sure, there are a few corner cases where FreeBSD could prove to be better, but in the real world people want reliability, performance, good software and hardware support, and the ability to run their OS for several years without having to keep upgrading. And right now, Linux rivals FreeBSD and beats it in many of those crucial areas. That’s simply the bottom line. They’re both great OSes, but Linux has the edge where it matters. 2003-12-26 3:43 pm We get it, you like linux. Good for you. Now stop posting the same flamebait every other post. 2003-12-26 3:47 pm Ok, my take on this… I think most people agree that both FreeBSD and Linux are good general purpose operating systems. No question about that. So for most tasks both FreeBSD and Linux get the job done. If however we choose to limit the discussion to the differences between the OSes I agree with Realist that Linux has the edge. My opinion I base on points like the following: – With regard to portability of the kernel, Linux has the advantage when considering the fact that Linux runs on more computer systems and processors than FreeBSD. – With regard to commercial support I believe it is a fact that Linux has more commercially provided hardware and software support. – With regard to general kernel scalability Linux also has the advantage. Linux works well on very small systems (remember IBM wrist watch?) to very large ones (Mainframes, Beowolf clusters, Big (Numa) boxes with hundreds of CPU’s) etc. – With regard to architectual components such as CPU schedulers, disk schedulers, VM, networking stacks etc. I don’t believe these are so fundamentally different that businesses and users in general should choose between these OSes based on them. For example. Ingo Molnar designs a new O(1) cpu scheduler for Linux. Soon after the FreeBSD guys design the ULE cpu scheduler based on it. With similarly the current Linux VM has been influenced by the FreeBSD VM. etc. etc. But if forced to choose I would give Linux the advantage for different reaons. For example: The network stack is complete free of the BKL (the Linux equivalent of Giant). The FreeBSD guys are still working on this. Also in Linux it it possible to do a open, read, write, close without ever hitting the BKL. Also the Linux kernel is preemptive. Making pretty low latencies possible is good hardware and drivers are in use. – With regard to RealTime support Linux has the advantage in this field as well. Consider two Linux based RealTime OSes. – With regard to virtualisation. Linux has UML (User mode linux) allowing one to run Linux as a process under Linux allow pretty cool applications. – With regard to ease of use and administration I believe Linux has the advantage. Not in the least because there literally exists hundreds of Linux distributions. On the one end you have the source based distributions. allowing you to do the equivalent of a “make world” and at the other end of the spectrum you distributions like Lindows and Suse providing a very nice “no hassels” desktop. From there you can move on to the Enterprise distributions from RedHat and others. Given the hundreds of Linux distributions I can but conclude that with Linux for business or pleasure you have more options. I could go on… but I think you get my point. FreeBSD only gives you on distribution and if you don’t like it you must role your own. Again. I don’t claim FreeBSD is a bad OS… My claim is that if you want to compare the two OSes _I_ find it pretty had to find technical or commercial _advantages_ for FreeBSD. The one point I believe FreeBSD has the “advantage” and I’m not sure it is an advantage… The BSD license. In some situations I can see this license has advantages. But this “advantage” to me, may be an important part of the reason why the BSD’s are progressing less rapidly that Linux. Anyway, this is just my opinion and I guess folks will disagree. 2003-12-26 6:24 pm Your logic doesn’t hold up whatsoever. If you really want the best tool for the job, you would not advocate using Linux. You’d advocate using Solaris, AIX, HPUX, and all the other real operating systems that scale circles around Linux instead. Also, you might be extremely happy that 2.6 is out, but only a clown would run their mission critical systems on that. I don’t care that the .6 stands for stable, the .0 thru .6 says “You’d be stupid to run this on a real server”. Finally, the only Linux distro I like is Debian and it receives no commercial support from any of the big players. It is extremely outdated. RedHat is a joke of an OS. It tries to be Windows with it’s GUI, but fails miserably at that. It also tries to be Solaris, but isn’t half as stable, scalable, and cost’s a whole lot more. And finally, you are a troll. 2003-12-26 6:26 pm “the .0 thru .6 says “You’d be stupid to run this on a real server”” I’m talking about 2.6.0 to at least 2.6.6 2003-12-26 7:07 pm “If you really want the best tool for the job, you would not advocate using Linux. You’d advocate using Solaris, AIX, HPUX, and all the other real operating systems that scale circles around Linux instead.” Wow, how vague can you get. You haven’t even said which “job” you’re talking about. It’s not always all about scalability; I only mentioned that as one reason for choosing Linux over FreeBSD. Do you really think that a small 2-CPU company webserver should be running AIX? Of course not. In that case, Linux _is_ the better tool for the job. Yes, Solaris, AIX et al are wise choices on vrey high-end hardware, but that’s not what this thread is about at all… “Also, you might be extremely happy that 2.6 is out, but only a clown would run their mission critical systems on that.” Erm, what on earth has that to do with anything? I never even mentioned running 2.6 on production boxes. There’s no need — 2.4 is very solid and still scales better than FreeBSD. On a similar note, some people have brought up FreeBSD 5.x as an example of catching up with Linux, but we could say exactly the same thing there! What’s your point? The fact is, FreeBSD users like to talk about their OS’s VM and with 2.6 Linux users have an even more refined VM, along with supremely good desktop performance. “Finally, the only Linux distro I like is Debian and it receives no commercial support from any of the big players.” Please, don’t shout when you can’t back it up. HP offers support for Debian — or are HP not a “big player”? “And finally, you are a troll.” Hehe. Really? There’s a BSD thread. Some BSD fanboys start ranting about nonexistant problems in Linux and spreading FUD and lies. A guy comes along, proves them to be incorrect and outlines very beneficial aspects of Linux that FreeBSD doesn’t match yet, in response, and he’s a troll? Never mind! 2003-12-26 8:16 pm LOL! One person pwned all of you BSD folks. 2003-12-26 8:20 pm i wonder how many times per day this ‘realist’ guy hits his refresh button looking for more people to argue with. 2003-12-26 9:25 pm All that I can see in the last comments is this: This news are about “Roadmap for FreeBSD 5.3”. I came here to read the comments about it and all that i can see is guys pro-linux from slashdot arguing the marvelous linux is. WTF is this??? Go back to yer LUG and yer slashdot, kid. This is a news phorum for serious people, not first-time-in-linux-kiddies trolls. 2003-12-26 11:47 pm Most of my boxes run FreeBSD. Why? Because in *MY* experience, FreeBSD performs better and is less buggy. I prefer BSD init scripts and configuration files because they suit *MY* needs better. That isn’t to say FreeBSD is for everyone. My friend down the road mainly uses Linux because it suits *HIS* needs better. Use what you feel more comfortable with, what you have good experiences with and what works for *YOU*. 2003-12-27 12:02 am What about security ?. I like linux It’s a great OS, maybe better than FreeBSD for desktop, maybe better performance for same task… but I prefer feel secure. FreeBSD model I think offer better quality and more audited code… this result in a more secure code. Admin tasks are fundamentals to maintain a secure system but with a lot of system while less problems (fix, patch, bug…) less work. I repeat, linux is a fantastic OS and I have used it for years and I will continue using but for now for my servers I will use *BSD. It’s my modest opinion. Happy new year. Saludos. 2003-12-27 12:36 am “1) Installing a server and leaving it in place for two to five years, applying patches remotely. Impossible with FreeBSD; you need a whole system update after 12 months Yeah, right. You claimed to write changes to FreeBSD kernel, yet you fail to recognize that FreeBSD has a binary upgrade system. Good job penis, you managed to make a joke out of yourself.” I find it laughable that you call me “penis” when you completely missed the point. FreeBSD doesn’t have a binary update system. There’s an unproven 3rd party system. And the main thing, as you quoted (and amazingly ignored) was the 12 months bit. After 12 months, your FreeBSD installation is unsupported. You either update the whole system, changing base packages in the process thus making it a different platform to target apps against, or stay with security bugs. Meanwhile, Debian offers two years of support. RHEL 5 years. Microsoft 5+ years. FreeBSD is well behind in this area, but of course you can’t admit that so instead you call me a “penis”. How mature. Just accept that FreeBSD doesn’t have half the popularity of Linux or Windows becayse 12 months of fixes per release is practically worthless in real-world deployments. You’ve made a joke out of yourself by ignoring that and going on a complete tangent… “Oh, yeah, 2.4 scales better than FreeBSD, right ? Wrong, it doesn’t and never will, get over with it.” That’s quite sad. You have a problem accepting the truth, and you need to get some friends or something. Here’s a little hint: John Baldwin, the man spearheading your SMPng operation, has said himself that FreeBSD is only at about Linux 2.2 performance. OK? 2.2. Wow, you look very weird now. “But of course, companies don’t want to run unsupported emulated software, no sir.” Once again, like a typical BSD zealot who doesn’t spend much time in the real world, you complelty misquote me. I never said that. I said that software _vendors_ do not offer the same extent of support for applications not running natively. Please, go out and look around and you’ll see this. “Not to mention some recent fights in Linux kernel camp.” Based on technical and social matters. It’s quite funny, because FreeBSD core has just thrown out Matthew Dillon from behind their closed, secretive list. Your OS is starting to suffer from lack of talented developers, and this won’t make things any better. “Name one driver that is specific and only available for GNU/Linux (don’t forget to mention which distro though).” You’re joking, aren’t you? And in 99% of cases the drivers are not remotely distro-specific. I know as soon as I list the x86 hardware Linux supports that FreeBSD doesn’t, you’ll come back with all sorts of nonsense about “not being relevant” etc. If you really believe what you’ve just said, go make a cross comparison between FreeBSD’s HARDWARE.TXT and Red Hat’s HCL. You’ll be very, very shocked. As for some of the other posts: “This is a news phorum for serious people, not first-time-in-linux-kiddies trolls.” Once again, when someone points out flaws in FreeBSD they’re labelled as a “first timer” and a troll. If you’d read up, you’d see that I’ve been running UNIX systems when you were still suckling on your mother’s teat. I’ve been a big FreeBSD fan, I’ve known what it’s like to feel you’re “different” and “better” than everyone running to Linux, but I’ve grown up and discovered that Linux can do everything FreeBSD can, but with better hardware support, improved performance on desktops and multi-CPU boxes, a wider software range, and much, MUCH longer fixes and updates for individual releases. Amazingly, absolutely nobody here has managed to refute those facts. Instead, they keep calling me a “loser” and a “troll”, when they know in their hearts that they can’t counter those facts. I at least expected some proper arguments, but then I realised — there’s nothing you can do. Instead of admitting that most Linux distros offer much longer support than FreeBSD, instead of accepting that Linux performs so much better on SMP boxes, instead of coming to terms with the fact that Linux has a broader and much more mature set of x86 drivers, and instead of agreeing that Linux can be just as reliable as FreeBSD, instead you choose to use personal insults. And that, my friends, is why Linux is being taken up. You have some worryingly anti-social attitudes, and it doesn’t help in you advocating the OS. Learn to live with the truth, use software that does the job better. 2003-12-27 12:39 am I am curious. I began playing with Linux about 5 years ago and have played with numerous distributions but now I use FreeBSD because it appears to be a much cleaner and better documented system and is all ‘housed under one roof’. My experience with Linux is that every distro puts stuff in different places and the documentation is quite poor. Compare this with FreeBSD’s docs which are beautifully written and presented. My questions is, is there a Linux distro that has documentation of similiar standards to FreeBSD? Can I have a link to it? I have thought of trying Xandros on my desktops and Debian on my server. I think Debian based Linux’s are the only ones I would bother with any more due to their slower development and therefore stability and also the installation of new software and dependencies (not an issue of FreeBSD) seems better than other distros. Thanks. 2003-12-27 12:48 am “What about security?” FreeBSD has a good security track record. But, as said, each release only has security fixes delivered for 12 months, so in the end you’re stuck unless you fancy a whole OS upgrade. Conversely, Debian GNU/Linux offers two years of fixes, and RHEL (and consequently the free Tao Linux / White Box etc.) offer a gargantuan FIVE years. This is very important to business, and note that the FreeBSD fans haven’t been able to counter this (without calling me a “penis” and so on). Businesses don’t want to keep targeting apps at changing platforms; they need something solid to base on. The proof is in the pudding. “Is there a Linux distro that has documentation of similiar standards to FreeBSD?” Yes. SUSE’s hard-copy documentation (750 pages upwards in the last release I tested) is a mammoth in-depth work. Red Hat’s professional docs are extremely detailed and thorough. Real facts. Unlike the hand-waving arguments we’re seeing elsewhere… 2003-12-27 1:36 am Yes, you are the true master. Even though you keep repeating your self with the whole ’12 of freebsd upgrades, pfff, i can have 5+ years with linux’. Ok, go ahead and run a 5 year old operating system. Good luck ^_^ Not that you can upgrade the whole OS in about 2 hours (over dialup) and be over with it. But i’m sure you’ll feel secure using that RH6.2 “John Baldwin, the man spearheading your SMPng operation, has said himself that FreeBSD is only at about Linux 2.2 performance…” In what year? 2000 ? “Once again, like a typical BSD zealot who doesn’t spend much time in the real world, you complelty misquote me” How so? I copy and pasted your statements, but of course, I’m evil. “I said that software _vendors_ do not offer the same extent of support for applications not running natively” And yet nothing is stopping Disney from using Photoshop on WINE. Nice try though. “Based on technical and social matters. It’s quite funny, because FreeBSD core has just thrown out Matthew Dillon from behind their closed, secretive list” Well, how about some more examples of all the people FreeBSD team is constatnly throwing out ? (btw. Dillon is still sending code buddy, his CVS access has been removed and that’s it) “You’re joking, aren’t you?… BLAH BLAH Yet, no facts and examples of specific drivers that you need so much. As to SuSE documentation: hahhahaaa SuSE documentation books (both user and administrator) are for people new to SuSE and unix concepts. While good for new people, they offer no value to any intermediate user. Not to mention that you have to pay for them. 2003-12-27 3:08 am I find it laughable that you call me “penis” when you completely missed the point. FreeBSD doesn’t have a binary update system. There’s an unproven 3rd party system. And the main thing, as you quoted (and amazingly ignored) was the 12 months bit. After 12 months, your FreeBSD installation is unsupported. You either update the whole system, changing base packages in the process thus making it a different platform to target apps against, or stay with security bugs. I try not to respond to you very often because you’re quite obviously an overzealous Linux troll, however I felt the need to correct the horrendously false information within your post. FreeBSD allows binary upgrades through /stand/sysinstall, the FreeBSD installer, by using the “Upgrade” option in the main menu. This has been part of the core functionality of the installer for over half a decade, and certainly isn’t an “unproven 3rd party system”, I’ve used it literally thousands of times across hundreds of machines. Please get your facts straight. 2003-12-27 3:10 am I would just like to comment on the benchmark above, that while FBSD reads much faster from a freshly mapped page, the creation of the page is so much greater. Take the following into account: Creation of 3000 pages mmap in microseconds (1/1000000 sec) L2.6: pretty darn near 0, call it 100000 O(1) L2.4: 1000000 O(n) FBSD-C: 2000000 O(n) (but scaling worse than linux 2.4) Read from first byte of mmaped pages (also at 3000) (In cycles converted to microseconds in parenthesis) L2.6: 2000 (2.2) L2.4: 2000 (2.2) FBSD: 100 (.11) Now FreeBSD is certainly faster initially reading, but compare this: Create + read: L2.6: 100000 + 2.2 = 100002.2 L2.4: 1000000 + 2.2 = 1000002.2 FBSD: 2000000 + .1 = 2000000.1 Now one might argue that FBSD might be faster reading in general, but the benchmarks above don’t test it (write some, and I will agree to compare) I would have to say that more benchmarks need to be done, reading from a random byte. Based on FreeBSD’s performance in the first, it may very well be that FreeBSD is much better than anything else at it. That’s the problem with benchmarks, sometimes one thing is highly out of proportion. Take OpenBSD, which like Linux 2.6 has a near 0 time for mmapping things, and scales O(1) add up the two 0+6000 (approx (bit high) average read from first byte)=6000 which on this beats linux 2.4 or FreeBSD. Though given the horrible time on reading the first byte, the subsequent ones will likely make the curve something like (assuming the next page takes just as long to access, an assumption that is highly unlikly) 0+6000*pages vs 2000000 +.1(pages) which evens out at about 333 pages (or on x86: 1.2MB (comparitivly 900000 pages for 2.6 & RBSD convergence, x86: 3.4GB) Benchmarks not helping again. Competetion is good: Linux 2.6 is ahead right now. Here’s hoping the Linux Kernel Devs don’t get lazy and let anyone else catch up, nor that the FreeBSD people don’t try to beat Linux soundly. 2003-12-27 5:40 am why can’t we stay on topic just once? 2003-12-27 5:58 am each release only has security fixes delivered for 12 months You mean each 4.1, 4.2, 4.3 … 4.8 and 4.9 releases? Then you would know that it isn’t hard to upgrade the system to the latest 4.9 even from the early releases like 4.1 or 4.2. It won’t that difficult to upgrade the system from 4.x to 5.x as well even tho you also need to rebuild/reinstall all the ports/packages. But again it will be more difficult to upgrade RHEL to the latest release. However at the moment, most companies may be using RH Linux 7, 8, or 9 and those versions won’t be supported anymore (7 & 8 from next year, and 9 from may next year, you know) There’s heck a lot of problems. Even when they purchase RHEL it won’t be easy to migrate all their servers from RH Linux to the ones they purchase. If they choose not to migrate there will be even more problems remain including security. You could say Debian will do, but company won’t consider it because big vendors won’t support Debian/Linux (Oracle, for example). For desktop use, it is too out of date. Businesses don’t want to keep targeting apps at changing platforms Yeah. So Oracle. Oracle desn’t want to keep targeting apps at changing platforms even among Linux distributions. Oracle would only support RH and SUSE at least in several years IMHO. And businesses won’t change their platforms from RH to SUSE or Debian. We don’t say they need to change their platforms from Linux to FreeBSD. I don’t have any idea why you say that? 2003-12-27 6:03 am It’s a pity that what could have been athread about the roadmap for FreeBSD 5.3 (which is something which interests me) degenerated into a Linux vs. BSD diatribe from the start. First that Keefer guy says obviously (to us, but not to everyone) false stuff and people go to correct it. Then that realist guy shows up and goes on feeding a campaign against we ‘zealots’, apparently not noticing that it is Linux which is OT here and not BSD. I don’t recall any of we ‘zealots’ showing up at Linux 2.x.y threads saying ten times in a row that Linux is very good but just worthless because there’s this thing called BSD which does everything better. Not because of its truth value but because it’s no relevance there. I take the blame for having taken the realist guy seriously and fueling his repetitions. Only later did I realise that when his fallacies are pointed out he just proceeds to ignore it and repeat himself again and again while claiming no one disproves him. And when he replies he choses what to reply to (as opposed to snipping the irrelevancies), and he choses that which is unimportant and easily attackable, silently dropping that which has some relevance. (Incidentally, you can’t have the support of RH with the tidiness of Slackware with all the pluses of Debian. It’s a pity, but that’s the way it is.) If someone still has the animus to begin a new thread on the roadmap… 2003-12-27 7:48 am Linux has maybe 10 times more developers than FreeBSD. It’s quite expected to see it new features in it faster… and I can guarantee you that it’s not because it was better at the beginning. It’s because of the hype surrounding it. It has been seen as being the sole competitor to Microsoft in the x86 world for quite a long time so I think it’s perfectly usual to see that it attracts more attention than the FreeBSD project. Like I said before, both projects are developed with a different philosophy. FreeBSD seems to focus on stability (developed slow and steady) while Linux is more on features to attract more users. Okay, we can argue the stability of BSD 5.2 vs L2.6.0 but I’m not aware of any BSD release that were messed up (unlike the many messes and oopsies that occured on the Linux 2.4.x branch). Anyway, use what is the best for you… Most desktops users will use Linux while many server admins will prefer BSD because it’s more mature (BSD is twice as old as Linux) and the userland is more unified. 2003-12-27 9:21 am Yes. SUSE’s hard-copy documentation (750 pages upwards in the last release I tested) is a mammoth in-depth work. Red Hat’s professional docs are extremely detailed and thoroug Suse is also a very expensive distro. I can buy a 2nd hand version of WinXP or Win2k for similiar $$$. As for RH, the docs you mention are free? 2003-12-27 2:01 pm Let’s just think about these comments some people have made: “Ok, go ahead and run a 5 year old operating system. Good luck ^_^ Not that you can upgrade the whole OS in about 2 hours (over dialup) and be over with it. But i’m sure you’ll feel secure using that RH6.2” Erm, RH 6.2 has alreaded been EOLed. You miss the point. And: “You mean each 4.1, 4.2, 4.3 … 4.8 and 4.9 releases? Then you would know that it isn’t hard to upgrade the system to the latest 4.9 even from the early releases like 4.1 or 4.2.” Look, it doesn’t matter if FreeBSD is so easy to upgrade that it sends out a lolipop through the CD drive tray. You’re both missing the issue that is crucial to businesses: long term stability. Witness the uproar when Red Hat changed their product lifespan for RHL to 12 months. Consider that while Microsoft products generally suck, they provide fixes for releases for years. This is supremely important to real world use, and it’s not down to easy-upgrading or anything like that. For example, a company develops an in-house application to run in its machines. It uses various components of the base system, along with “external” software (ie Ports or RPMs or .debs etc.). If the company chooses to go with FreeBSD, and installs 4.9, then in 12 months they have to upgrade the system or take risks with security and undiscovered bugs. But if they upgrade, they have to do massive testing as there will be slight differences in the kernel, C library, userland and packages such as Perl, Apache, and loads of others. Conversely, if the company deploys Debian, they can run their application for at least two years on the same machine with backported security and stability fixes. With RHEL, they can leave it running for 5 years, knowing it’ll be secure. And this saves so much time testing. Companies can’t afford to keep re-testing their applications year after year, and that’s why Microsoft got flack for EOLing NT4. All users had to shift to 2k or XP, and while the upgrade isn’t hard, re-testing and fixing their apps is a very lengthy and tedious process in mission-critical scenarios. So _this_ is the point. And I’m not attacking FreeBSD because of it; they have a much smaller community so it’s hard for them to offer longer lifespans for each release. But their 12-months-before-EOL is one reason why Debian and RHEL are being taken up so much more. Look around the Net and you’ll see this point brought up time and time again. Linux has an enormous advantage here. “and I can guarantee you that it’s not because it was better at the beginning. It’s because of the hype surrounding it.” Bit chicken-and-egg, though. When it was a level playing field and both OSes were essentially geek projects, why did Linux take off? It wasn’t hype _then_, because neither OS was hyped. “First that Keefer guy says obviously (to us, but not to everyone) false stuff and people go to correct it. Then that realist guy shows up and goes on feeding a campaign against we ‘zealots'” So your correct Keefer’s lies, and when I correct your lies (Linux is just hype, it’s slower than FreeBSD, it’s less stable) I’m suddenly on a campaign? “And when he replies he choses what to reply to” Because I’m not interested in debating the merits of “make world” or a “tiny bit faster TCP/IP stack”. I’m talking about real-world Linux usage; the reasons Linux is being widely adopted. And hardware support, lengthy product lifespans, wide software range etc. etc. are simply far more important. It’s well known that people respond badly when they feel threatened. If someone is happy with their choice, they don’t feel the need to denigrate others. Which is intriguing, as I’ve always said that FreeBSD is a good OS with some decent features. However, the BSDers are _obsessed_ with Linux, constantly trying to pick fault with it and saying things which simply aren’t true, and then resorting to personal insults when someone comes back at them with hard facts. I keep re-saving this thread to demonstrate the problems with FreeBSD “advocacy” in future 🙂 Me: “Don’t keep spreading the idea that Linux is unstable and slow, when it’s not. Also, hype isn’t why it’s popular, it’s because of hardware, apps and lifespan – things that businesses need. FreeBSD is still a good OS though!” You: “You’re a zealot, a penis, a first-time Linux user, clueless, flamebaiter and troll. make world and wireless drivers.” Who’s being rational, fair and decent, with genuine arguments? Who’s using FUD, insults and hand-waving? 2003-12-27 4:13 pm Can’t you “Realist” and other GNU/Linux advocates realize that we’re not the least interested in what you think about Linux? It’s no sense in arguing at all with you since your only interest is to decrease diversity and throw in flamebait because you have nothing better to do. Why is it so darn important to you that every soul on this planet runs Linux??? We are fully aware that Linux exists, that we can download it and that we can install a whole bunch of things which you probably think is very “l33t”. Now figure this one out, why is it that BSD users rarely have to advocate how brilliant their OS is but just let the OS itself show off, while Linux users must speak so phreaking much of it all the time and not deliver? I’ve tried Linux, I wasn’t satisfied… why should I try again? I got promised a whole bunch of things that time, and get promises again and again and again… YOU Advocates are the very reason to why I hate Linux, will never use Linux and will never advice anyone to use it… because all info you give is UNRELIABLE… Using BSD, you know what you get, for better or worse, I don’t care, as long as I know what I get! 2003-12-27 4:25 pm This isn’t advocacy. This is defending Linux from the FUD and lies that are being spread here by BSD users. You would do exactly the same if you found your OS was being unfairly slated. “Why is it so darn important to you that every soul on this planet runs Linux?” Did I say that? No. For the zillionth time, I don’t care what people run. I’m just asserting the truth and facts, not hand-waving. “why is it that BSD users rarely have to advocate how brilliant their OS is” Did you read this thread? It’s full of BSD users trying to diss Linux and promote their OS with inaccuracies. Now go look at BSD forums and mailing lists around the Net. Every other post is a mention of how Linux is just “hype” and “worse”. Many BSD users are obsessed with Linux. “YOU Advocates are the very reason to why I hate Linux” Wow. Remind me never to employ you. You “hate” an OS because of what some people say? Personally I prefer to make my judgement on technical matters, but whatever floats your boat… “I got promised a whole bunch of things that time” Linux users keep getting promised by BSD users “better performance” and “more stability” etc. When they find that’s untrue, and see areas in which FreeBSD is weaker,they get turned away for good. The BSD community is, right now, its own worst enemy. 2003-12-27 4:57 pm I won’t be adding any more to this thread; I’ve defended Linux from the FUD and used genuine real-world arguments instead of personal insults and hand-waving. BSD users, feel free to dismiss Linux’s popularity as “hype”. That’s your choice. Just remember that you won’t learn anything from denial, nor will your OS get any better by failing to see the aspects in which other systems have improved. It’s not hype. The points I mentioned — hardware and software support, long release lifetimes — are very important in serious work. Hundreds of thousands of technically knowledgable people the world over are deploying Linux because it offers these benefits; dismissing it all as “hype” means nobody learns anything. EOT! 2003-12-27 7:09 pm “BSD users, feel free to dismiss Linux’s popularity as “hype”.” Linux is hype. Why else would IBM start selling Linux on it’s high end machine when their own OS owns Linux? It doesn’t make any sense whatsoever. As with your 5 year support plan from RedHat… well there are a lot of companies that do not want to pay $2000 rental fee, especially per computer. There are also a lot of company that were burnt by RedHat when they inexplainably dropped support of their RedHat standard product, including about 10 servers in our company, causing mass panic and mass abandonment. In all the years we’ve had our FreeBSD boxes, we haven’t had any disappointments. FreeBSD is free and you can install it on an unlimited number of computers. Also, in all the years we’ve use it, it hasn’t change it’s policy, screwing 99% of their customer base, to make a shiny nickel. In my company, I can’t justify paying $2000 / server / year so that I don’t have to upgrade every twelve months, a 15 minute job / machine. And that is a fair estimate in time. I choose a very minimal set of applications to install on the FreeBSD box (better security, easier managability), so it takes at most 30 minutes to upgrade FreeBSD. I don’t have to worry about breaking anything either because I do not run third party commercial applications on it. Now what FreeBSD offers is very simple. It offers an open platform that supports POSTGreSQL, Apache, BIND, Postfix/Sendmail, FTP, SSH, Samba, and a host of other very usefull services, and it offers them reliably. These are all the services I want to run on a free OS. It might not scale well to 64 cpu’s but it does a pretty good job scaling to 4 cpu’s. Even with this scalability, most of the computers I use it on, are uniprocessor/dual processor machines. In the very near future it would easily scale to 16 cpu’s. But I have a feeling that I will still be using uniprocessor/dual processor machines. The same is true with all the Linux machines. Finally, if I wanted to run commercial software, such as Oracle, I wouldn’t choose RedHat. On the high end, I would choose Solaris. That way I can actually purchase enterprise ready hardware and enterprise ready scalability. On the middle/low end, I would choose Windows. Windows only costs about $1200 (that is a one time payment fee versus $2000 /year for RedHat) and offers equivalent scalability and reliability to Linux. 2003-12-27 7:47 pm OMG this thread is such a joke. I guess a flamewar is the only reason people post in BSD threads. LOL j/k Anyways 5.3 seems like its going to be pretty sweet. 2003-12-27 8:23 pm Now go look at BSD forums and mailing lists around the Net. Every other post is a mention of how Linux is just “hype” and “worse”. Many BSD users are obsessed with Linux. ever read slashdot’s BSD section? it’s the same thing as you describe here. I uderstand that such things come from both sides but what I don’t understand is why you’d bother posting your “corrections” at all. I don’t usually post on Linux threads about BSD (I’m sure I have once or twice but only for a laugh, you seem so serious about it). Well whatever you do with your time is your business I suppose. Back on topic, I can’t wait for 5.3. It is an exciting time for BSD users. jared 2003-12-27 8:41 pm It seems this thread is wishes to compare two OS’es; FreeBSD and Linux. FreeBSD is an OS but Linux is not, that is why it is more correct to say GNU/Linux, of course we all know that. Mr. Realist is drawing attention to features of Linux that he thinks gives it the edge over FreeBSD. But the features are specific to a particular distro. He claims that Linux is easier to administrate, as tidy as BSD, longer support for updates, and better hardware support. I think the main features are: Feature OS/Distro apt-get Debian tidiness Slackware Long updates RHED Hardware support All/Linux kernel No single Linux distro combines all these features. When setting up a server one cannot install all these distros to get the different features. One has to choose. Another thing, why does every thread bring up that scalability study to claim that Linux outperforms FreeBSD? The study isn’t even about performance but rather scalability. It makes no sense. Looking forward to 5.2. 2003-12-27 10:21 pm What is it that is common to 90% of OS flamewars? That one of the sides is pro-Linux. Now go and ask yourselves: where are most of those flamewars taking place? On a non-Linux section. Then ask yourselves: what are pro-Linux people doing on a non-Linux section? Just browsing? Ok. Flaming others? Well, go away. Really, most people on non-Linux sections go to a Linux section when they want to hear about Linux, which may be often, seldom or never even. Most of Linux users, also, don’t seem to realise that whoever uses another OS (Windows or Mac OS X excepted) most probably has already tried Linux. It’s the Linux users that use the ‘mainstream alternative’ OS, not the other way round, so it’s kind of pointless to evangelise it. 2003-12-28 6:16 am I don’t know why realist repeatedly says “hardware and software support, long release lifetimes”. As what Zhang posted earlier, no one specific distro features all of them pointed by realist. Most if not all of commercial softwares like Oracle are only supported on Red Hat and/or SUSE. But Red Hat is mess (not tidy like Slackware). There is apt-get for RPM but is it that good as the orignal apt-get on debian? And Red Hat’s RHEL is too expensive for most companies using Linux. $2000/server/year is totally overpriced IMHO. It is not affordable especially for companies running a server farm with.. like several hundred or thousands systems. If the company chooses to go with FreeBSD, and installs 4.9, then in 12 months they have to upgrade the system or take risks with security and undiscovered bugs. Why does he have to say ‘installs 4.9’ and ‘in 12 months’? Well, let’s compare it with Red Hat. Say just before Red Hat announced the EOL of 7,8 and 9 Simply say it was October. Company A chose to run latest RHL 9, and company B chose to run FreeBSD and it was 4.8. Which company gets longer lifespan on their systems? Company A gets around half a year!! where as company B gets at least a year+alpha from the beginning of their business. And also company B can upgrade their systems to 5-stable when it is released using sysinstall, or cvsuping and make world. even though it takes a bit long it works and they get longer lifespan. I wish 5.3 to release ASAP so that we can benchmark it meaningfuly with linux 2.6. Anyway, my choice of OS is FreeBSD for its stability and performance. I couldn’t feel better with Linux that made me to go back to Windows within 1 month every time.(and go to Linux again 1 month later. it was a sort of cycle) Now I am happy with FreeBSD. Not going back to Windows. Sorry if my choice of words wasn’t good enough. A Happy New Year to ALL! 2003-12-28 1:35 pm Diversity is important. I think that a bigger development community (as in the case of linux) naturally can develop implements faster (sorry for bad english). So, in the long run, maybe linux will “outrun” FreeBSD in some way. It would be sad though. Before you all start to post replys, I use gentoo, FreeBSD, OpenBSD and W2K. Every system is used where it fits best. 2003-12-28 2:52 pm “And Red Hat’s RHEL is too expensive for most companies using Linux. $2000/server/year is totally overpriced IMHO.” (Yup, I said I wouldn’t post again and I don’t mean to, but this requires some clarification.) I agree that it’s expensive. But there are alternatives based on RHEL such as Tao Linux and White Box Linux. Both allow you to use Red Hat’s errata SRPMs, so you can still run a Linux machine securely and reliably for 5 years without paying any money. It can’t be stressed enough how important this is to businesses. They don’t want to upgrade machines every 12 months. This is one reason Microsoft is successful; long-term support. “Linux is hype.” You insult hundreds of thousands of very technologically knowledgable people around the world deploying Linux. They’re doing it because it works, it doesn’t need to be upgraded every 12 months, and it has a broader range of drivers. Those are real-world benefits, not hype. My last word, to the people on here who’ve said they “hate” Linux and think it’s “hype”, bear in mind that the software you use — Apache, Sendmail, XFree86, GCC, GNOME, KDE etc. — is being developed primarily by Linux users and companies. Linux is driving the rapid development of these projects; much of the great software available in FreeBSD is mature and featureful because of work by Linux users and corporations adding, testing and tuning code. Whatever your feelings about Linux, the work of IBM, Red Hat, SUSE, Novell, Intel and thousands of developers is making the robust, highly featured and respected software today. Go back to XFree86 3.x, GCC 2.7, and so on if you want software that hasn’t been “tainted” by the massive Linux effect. Bye!