Slashdot Mirror
OpenSolaris Or FreeBSD?
Norsefire writes "I am in quite a predicament. I decided a while back to branch out and use a new operating system (currently running Debian). After a bit of searching (trying Gentoo, Gobo and Arch along the way), I decided to use something that isn't Linux. Long story short: I narrowed the choices down to OpenSolaris and FreeBSD, but now I'm stuck. OpenSolaris is commercially backed by Sun, has nice enterprise-y tools in the default install, and best of all, a mature implementation of ZFS. FreeBSD is backed by a foundation, has a minimal default install and a rather new (but recently improved in the 8.0 release) implementation of ZFS, however it offers the Ports Collection (I quite like the performance boost due to compiling from source, no matter how small it might be) and a bigger community than OpenSolaris. That is just a minimal mention of the differences. I would be interested to see what the Slashdot community thinks of these two operating systems."
Rather than playing with just another un*x clone, try something like Haiku or FreeVMS or my personal favourite Plan 9
Andy Warhol got it right / Everybody gets the limelight
Andy Warhol got it wrong / Fifteen minutes is too long.
When it comes to things like flash, acroread, nvidia drivers, fluendo (multimedia plugins, DVD Player), skype etc being supported, having the commercial entity behind OpenSolaris does seem to help...I think behind the scenes Sun offer some sort of incentive to these companies to support OpenSolaris. I do like that FreeBSD is backed by a foundation though, it is much more reassuring to an open source project to know that its backing entity wont dump them the next day.
Looks like this is "just for fun" or to learn new, interesting things. A good reason, if you ask me.
Having used both briefly, I can't think of a good answer other than "try both" or "flip a coin" - neither is better or more interesting than the other and both are different from Linux in many subtle ways, enough to force you to learn something, and to cause that funny feeling when you perform some learned, almost mechanical tasks as if it were Linux and almost forget it isn't, when suddenly something unexpected happens (as in, a command having completely different output formatting or existing under a different name, or a subtle difference in directory structure, not a spurious rm -rf /, hopefully).
This is Slashdot. Common sense is futile. You will be modded down.
I assume you are looking for a server. If it's for a desktop, more users and software help a lot. Although BSD and Solaris are more reliable indeed, the intricately, meticulously designed user-oriented design interface of Linux provides details and config files enough to entertain for generations. I have never tried out GnuStep, however an open source nextstep-like interface seems promising.
Build your own energy sources from scratch. http://otherpower.com/
Make a VM of each system and see what you like. The other question is what do you want to do with your system? Run it on your laptop? Use it as a web server? A directory server? Or something else?
This is question is like being asked by a computer illiterate user "What kind of computer should I get?" I always ask "Well what do you want to do? If you want to surf the web, maybe type a paper or two, get a netbook, if you want to play games, get a desktop, if you need to carry it to school or work..." It all depends on what will best preform the functions you're looking for.
If your goal is to learn, try both.
I only have my own anecdote about this, but I kind of like it.
Back around '00, I had several computers (I still do, but that's beside the point). I had my main desktop, and I had this nice old Pentium 200. I also had a TV-card (Hauppage, I think). If I tried using the TV-card on my main desktop, it would be hellishly slow for doing other things. In addition to some of my screen being covered by the TV-window, of course.
So, I installed the Hauppage card in the P200, which was running stock FreeBSD. It worked, sort of, but the machine was almost unusable for other things.
I tuned the kernel, fiddled with compiler flags, and remade the world.
And what do you think? The entire machine went from lurching slow to usable, while displaying TV. It was the "little extra boost" that was needed.
Now, of course, I don't think it would be of much use to me in most cases these days - as machines have grown so extremely much faster since back then. But, it's the story I tell whenever people ask about performance boosts from recompiling everything.
"Rune Kristian Viken" - http://www.nwo.no - arca
I've used both as my primary desktop (each for a few months) and if you want to try something new, go with FreeBSD. OpenSolaris felt like GNU/Solaris, which it mostly is, with a few Sun coded things (I think it was libc and a few more of the libraries). FreeBSD was all about fine control: I found myself wanting to recompile the kernel and playing with rc scripts and asking my OpenBSD-using friend so many questions he demanded I switch to Linux:-D
Plus, when you've spent a whole night figuring out why KDE won't compile correctly on FreeBSD....it feels good, like you've accomplished something.
I'd have to agree here. Although FreeBSD's ZFS support is getting quite good now. I'm using it on a production system and it hasn't let me down. It even saved my bacon a couple of times (yay, ZFS snapshots). I guess it depends on what you want to do. Both have strong features. OpenSolaris has Crossbow, but FreeBSD will have vimage soon. Both have Dtrace and ZFS. Solaris has zones, FreeBSD has jails. But I think FreeBSD is easier to tinker around with (personal opinion).
Does ZFS on FreeBSD still suffer from random kernel panics when it gets low on memory?
I'm particularly referring to this bit of documentation:
To use ZFS, at least 1GB of memory is recommended (for all architectures) but more is helpful as ZFS needs *lots* of memory. Depending on your workload, it may be possible to use ZFS on systems with less memory, but it requires careful tuning to avoid panics from memory exhaustion in the kernel.
Yeah, kernel infrastructure that can't cope with running out of memory. That fills me with confidence. Particularly I've run ZFS on OpenSolaris on a 48MB Pentium laptop and it coped fine.
Unfortunately the FreeBSD ZFS pages are a wiki, which means they're badly organised and out of date. I have no idea when the above was written or whether it's still valid. Does anyone know?
http://en.wikipedia.org/wiki/XOR_swap_algorithm
This maybe wasn't the best example since XOR swaps are rarely useful anyway. I suspect that other things like word (mis)alignment and varying cache miss costs may be a factor for different processors.
Gentoo claims that picking e.g. core2 over nocona can boost performance by 15% (which seems a bit much to me), so picking the right x86_64 variant is still something that is considered. Not something I worry about though, unless I am compiling from source anyway.
Why would code compiled on your system run any faster than the same code on someone else's system?
Because many pre-compiled packages use conservative optimization flags and may lack specific code paths for certain processors and instruction sets. They might also have chosen a compiler which doesn't produce the fastest code around. I'm not sure how it stands today, but a few years back, ICC produced code up to 30% faster than GCC or MSVC.
The difference all depends on the type of application of course. Overall, you might only see a performance difference of 1-5%, but for specific parts of the application, performance increase may be anywhere between 10 to 200%.
Last, compiling yourself also means you can choose what gets compiled and what not. Which in turns reduces diskspace and memory usage of the executable and may increase security and performance a bit. For things like Kernels and such, you need to compile it yourself if you want support for specific things (ALTQ for PF under FreeBSD for instance).
It only takes one man to change the Wisdom of the Crowd to Tyranny of the Masses.
(Yes, I have used Gentoo. For several years. I concluded at the end that the amount of work was greater than the benefit.)
Me too. I love Gentoo, and think it's pretty much as close to my perfect distro as possible. Gentoo Hardened is brilliant.
However, if you do what I do, and only update packages that have security issues, you'll find that suddenly one day, your profile has expired, and packages you need to bring it up to date have entered and left portage, meaning that you have to jump through hoops just to get Python working enough to update.
Say anything about this, and you get the statement "Just do emerge world every night", which is stupid for a production server.
I much prefer Gentoo to Ubuntu or Debian (and nothing to do with speed (claimed or otherwise)), but my current host? Ubuntu 9.04.
Get your own free personal location tracker
I'm using Debian stable right now as the solution for my particular requirements (development desktop that's a good Xen Dom0), but I'd much rather be using a BSD (the first machine I bare metaled was BSD 2.x onto a PDP-11/44 in 1981 (sic)) or Solaris (it took me most of a decade, but I eventually got over their switch to AT&T :-).
The big problems with FreeBSD when I made my decision were no Dom0 support and an immature ZFS, and the problem I've always had with Solaris is solid mass storage device driver support, at least for vaguely affordable controllers that don't require a PCI-X bus. E.g. when I last checked nVidia SATA chipset support was iffy (which was odd since a classic workstation they shipped had a rebadged Tyan motherboard with a nVidia chipset; I've got two of those Tyans in prodution and they're rock solid ... with Windows XP :-( hey, I'm not willing to put my parents on Linux or whatever quite yet )).
This may have improved since then, but be sure to check for problems in the field.
What makes you want to blow away something you're already running & comfortable with? You give no reason for switching away from Debian.
Suggestions:
- For Linux, Debian is pretty much the granddaddy, and can likely be wrangled to do whatever you want. You seem the explorative type. If you're comfortable with Debian, figure out how to do whatever it is you're interested in on Debian and get on with it. Changing distros won't change your life.
- For other OSs, you're blessed to live in the age where you can just grab virtualbox, fire up a VM of whatever it is you wan to play with, and fiddle with it. When I was messing with all this I had 5 crappy old noisy minitower PCs around my desk (and a NeXT on top of it, which was what I actually used as my workstation becuase it Just Worked). If you're really really impressed by something that you've monkeyed with in your VMs for a while, switch to it if you really want to, but honestly in ISP and hosting type shops Debian is what I see most.
- It sounds like you want slowlaris or FreeBSD just to get ZFS, presumably because you have an ever-expanding collection of media, pr0n, und w4r3z and want to be able to just add disks to your storage pool on the fly and all the other spiffy stuff that ZFS does. If you want to kick the tires on a new filesystem technology, may I suggest that you grab the latest iso release of DragonFlyBSD and check out HAMMER? It's really a lot simpler to use than ZFS, and personally I feel it's really designed The Right Way.
- If you really want a challenge, get a Mac (or buy yourself Snow Leopard and make yourself a hackintosh) and learn how to use the powerful and complicated tools that make Mac OS X Server work. Things are very different from the way other unixen do things, and I find messing with them and learning how they work to be very satisfying.
I expect some BSDs to flourish as well, but if you think that'll happen because the GPL/LGPL stack is somehow shunned by the commercial players... well, all I can say is that you are the one with blinders on.
Copyleft is successful now because companies see value in contributing to copyleft software. You can keep trash talking all you want, that doesn't change the reality.
Trying to pin the non-success of SFU on the open source community is especialy rich...
I came from a SunOS background but used Linux based distributions at home (Slaskware was the easiest at the time).
I the tried NetBSD and FreeBSD and they were okay, I found general responsiveness felt good, not necessarily faster, but more consistant, this was years before low lateny linux kernel.
After about 9-12 months, I realised I was spending a lot of my time just trying to get iBCS, Wine and Linux compatibility working so I could be productive. I realised I wasn't gaining anything from running FreeBSD
and was struggling to make it work like a Linux based desktop OS. As a server I favoured Solaris anyway.
I'm quite happy with both OpenSolaris and FreeBSD as desktops, as well as servers.
You didn't specify what your primary goals are for the system in question-- if you're looking for a general purpose web surfing/light development machine, OpenSolaris should be fine for you-- as long as you have at least a gigabyte of memory and a moderately fast processor.
FreeBSD's a lot less resource intensive in my experience-- I'm currently supporting two sites that still have Pentium III/600-based servers with uptimes approaching a year each. (Last reboot for each was due to a multi-day power outage.)
If you have VirtualBox installed, give both FreeBSD and OpenSolaris a whirl, see what you think.
Therefore if we are to restrict our options to OpenBSD and FreeBSD i would lean towards FreeBSD simply due to the large no. of apps available through ports.Also i believe driver compatibility is a little better in FreeBSD, especially recently with nvidia cards.
FreeBSD only had NVidia on i386 kernels at least when I tried it on the desktop and quit about 2 months ago. You have the open source driver with works, but for decent multi head on a new model card you still need the closed source driver. OpenBSD has similar and (in my case) sometimes better hardware support. The Intel wireless card in my Dull laptop is supported on OpenBSD out of the box but FreeBSD still required me (at 8.0-RC1) to download a driver and munge with boot options to make it happen. Doable, but mildly annoying since it is the essentially the same driver with an extra PCI ID added to it to let it use the card.
If you're building a kick-ass server then the choice is up to you. You can't go wrong with Slowlaris or *BSD. I like FreeBSD because it has jails. They take a little getting used to but they are great. It's particularly useful to be able to give people root in a jail to admin something and know they can't actually root any of the other jails or the host. Solaris has Zones. Linux has a suite of patches that can do jails but it's not mainstream yet, and I wouldn't trust it as far as I could throw it if I were trying to attach a patched kernel to a current distro.
Solaris and FreeBSD have ZFS. Both are stable. Solaris has the slightly more mature support, but I've never seen FreeBSD lose data or kernel panic on me because of ZFS. There's a LOT of advantages to using ZFS. Quite a few are met with LVM on Linux or dynamic disks on Windows, but not all.
OpenBSD is going to be more secure out of the box, until you start installing from ports or packages.
Solaris is heavy. The default install was massive last time I tried it and it took forever to boot. Linux is even worse on size but fast to boot now (Ubuntu and Fedora at least have made huge advances in their latest releases). FreeBSD and OpenBSD can be shoehorned into very little space if needed without resorting to rolling your own distro, which can be advantageous.
For hardware support, Linux is generally better than the alternatives. If you only have very new hardware then Windows 7 is going to have better support.
Right; it should be clear as mud now that every system has its own advantages and disadvantages. It's like asking "should I choose a Porsche or a VW" (Ok since VW owns half of Porsche now...) Horses for courses as they say. You'd test drive the cars if you were looking to make sure they met your needs. Try the different operating systems on a sacrificial machine or VM. Stick with the choice that you feel comfortable with.
I drink to make other people interesting!
The user clearly doesn't want some 'dual-boot' system to run a server operating system ...
The OP fails to mention what he actually wants the OS for, other than mentioning that he likes "enterprisy" solutions. It's difficult to offer useful suggestions when all he wants is something that isn't Linux, but can't (or won't) articulate why.
"Raising the bar for competition" is exactly what BSD does, and just look at how much it helped Apple and Google to finally start offering some serious challenges to Microsoft. GPL actually helps the biggest commercial player (i.e. Microsoft) retain their position, because they can afford to put huge amounts of money into R&D while their would-be competitors cannot. Sure, GPL will eventually squeeze them out of some over-saturated fields, but Microsoft will always be able to invest in other things where it can make a profit: business services, hardware, and so on.
Calling me a "Microsoft apologist" does not change the basic economic facts. And, in reality, the only type of "monopoly" that has ever existed in human history was the regional hegemony of government force - everything else is subject to perpetual change.
Its a good point there. I work with a Microsoft shop but somewhere along the line they decided to support Oracle databases running on Redhat. Since then, I'd say the majority of our customers running Oracle have plumped for Redhat (the others won't until they upgrade). I wonder if Oracle will be trying to scrap this in favour of Solaris (not OpenSolaris surely) and charge lots of money, but I doubt any of them will migrate - migrating from Windows to Redhat makes a lot of sense, even if the cost is roughly the same overall. Migrating to a more expensive Solaris OS probably won't.
Sun made itself as irrelevant as anyone else, they were the Apple of the server world, selling overpriced hardware that wasn't much good compared to the equivalent you could get from IBM (we did this, 2 pedestal servers, the IBM was 4x the computing power, cheaper, and a much better build quality). It wasn't Linux that killed them.
So what if the case was settled in 1993? As a result of the case, the AT&T code was removed from free BSD distributions. FreeBSD didn't have a cleaned-up release version until 1995.
Of course they would. A company wants more customers. So Interix "was created solely for the reason of destroying UNIX" in the same way that any company tries to make their products better than competing products.
I suspect people contribute to these kinds of projects because they use them, not for the only benefit of helping "the community". People do use computers for useful work, from time to time.
After the BSD lawsuit, the open source BSD distributions were rewritten without infringing code. This took some time; the non infringing version of FreeBSD wasn't released until 1995.
The SCO lawsuit had no effect on Linux because it was immediately recognized as nonsense from the beginning.
In the 90s, Minix couldn't even be freely distributed. As a useful operating system, Minix didn't compare to Linux or BSD back then.
Netscape was closed source and commercial for a long time. By the time the Mozilla project was started/Netscape was open sourced, IE (another closed source browser) had already gained significant market share and Mosaic had long been irrelevant.
The source is always available, yes - and if the feature is useful to others, and someone else has an interest to put that feature in the mainline Linux kernel, they can. Otherwise, the code will just get stale.
Companies are hardly "in" the community if they do nothing other than honor the GPL obligation to release the source. The criteria I am using: Do they contribute their useful modifications as patches to the original project or participate in the communities of the projects they use?
I'd say that statistically, ZFS is now safe to use from the point of memory allocation failures simply because the number of user reports to it has fallen off dramatically after the new version and resource limit patches got in (which was significantly before 8.0-release so there was plenty of time to observe the effects).
-- Sig down
As a long-term Linux guy (since 1995) I think ZFS integration with Samba, iSCSI Targets, and Zones makes OpenSolaris relevant to me. I am now trying to learn Opensolaris so I can use these on a SOHO server. Sure in a year or two BTRFS may have RAID5-like redundancy, caching and intent logging on SSD, and these features, but OpenSolaris/ZFS has them now. I definitely won't be running any solaris on my netbook (kubuntu), laptop (WinXP), or Macs though.
I think if the opensolaris community can produce a variant that makes it easy for a less-than-elite user to set up a server with a samba share, some iSCSI targets (for time machine, aperture library, or whatever), and possibly an IMAP server, they can greatly increase the pool to whom they are relevant. Auto-magic HDD management like drobo would help too.
I love running DOS15 on PDP-15 SIMH emulator; the installation was almost an adventure, but eventually got there. As far as the OpenVMS goes, I run it from time to time on a OpenSolaris host (two dual Opterons) and it is faster emulated than on any real hardware I have ever run it on (well, the fastest VAX m/c i've ever used was an 8700, started on a 730...).
Gone are the days...
opensolaris does have a packaging system but last I checked it did not have a whole lot of different software packages. The ports collection has everything you could want. If you are going to go with opensolaris check whether the software you plan on using is there or be willing
to compile and/or package your own.
The freebsd zfs implentation is pretty mature now in 8.0 but it's not the newest version and it doesn't have all the latest features that opensolaris does. This isn't neccessarily a bad thing though, look at the differences and see if you actually need those newer features now.
What are you going to use this box for? Overall it really comes down to looking at what requirements you have for the box and looking at which of the 2 meets those requirements the best. If you can't decide then just try them both and see which one you like better. Thats what I did and I settled on freebsd.