Where Does NetBSD Fit In?

NetBSD Fan writes "KernelTrap offers a fascinating summary of the recent 2004 Annual NetBSD Group Meeting. Included is an introduction by NetBSD foundation president Christos Zoulas discussing NetBSD's relevance in light of competition from well known operating systems such as Linux and Windows which he acknowledges 'both offer more features than we do, and they have behind them the resources of very large commercial organizations.' He also talks about FreeBSD, OpenBSD, Solaris and Darwin, ultimately concluding that they all are facing their own serious challenges, and that plenty of opportunities remain for NetBSD. The NetBSD project recently released NetBSD 2.0."

Comments

Everyone knows

[Anonymous Coward]

FreeBSD is the popular one
OpenBSD is the secure one
and NetBSD is the one that'll run on my electric toothbrush

On the firewall

[Anonymous Coward]

NetBSD my firewall OS of choice.

Re:On the firewall

[vstanescu]

You are just wrong.. if there is something that you really know, probably you will be able to configure that operating system to be more secure than a default-secure operating system that you have no knowledge about. So his choice for what he knew is the best, not the security advertising about openbsd.

it fits on my old SPARC

[ChipMonk]

In fact, the latest release of NetBSD fits better, and runs faster, than the Solaris of 4 years ago.

runs on old and rare archs

[lanc]

NetBSD runs on 17 CPU architectures. Can you count up 17?

NetBSD will be the OS what you can always use on your old boxes, when you don't get running anything else on them.

Re:runs on old and rare archs

[setagllib]

You can't do that with Linux either. The name aside, the Linux you run on one arch is not the Linux on another arch. This is especially true if you have an architecture Linux' corporate sponsors don't care about (hint: Linux is all about corporate sponsorship. If you don't believe me, read the changelogs and notice that 99% of the submissions are from employees of big corporations).

Some distros (e.g. Debian) smooth this out, and some make it a nightmare (e.g. Gentoo), but until there's a GNU/Linux distribution that is consistent across all archs in source AND binary deployment and a kernel that contains all architecture fixes at once and keeps up with mainline development, Linux will not be as easy to 'jump between' archs as NetBSD.

NetBSD has the same build procedure on any architecture, using the same headers, sources, and resulting software (sans tools that only make sense one some archs). THAT is the same OS on every arch. The same name of kernel on every arch doesn't even compete. Being able to cross-compile consistently is a great bonus too, but this sometimes breaks down if you're following a development branch. Not all archs have an installer, but those that do appear to have the SAME installer, with extra functionality for those architectures needed.

My example of where even Debian does not have this: SGI MIPS. It does NOT automatically handle the SGI Volume Header (hint: NetBSD does, and installs its own bootloader), nor even tell you what to do: you have to figure out how to use the fdisk-like editor and hope you left enough space for arcsboot or your kernel. On some machines which have very tiny hard drives and need all the space they can get, Debian's way leaves a lot of user calculation to be done. You could call this "well if you're incompetent then don't use Linux", but then that's supporting my argument: NetBSD is an OS for all archs, Linux is a kernel that got ported to some archs at some point in time, and if they're not used by corporate sponsors you can kiss them good bye. Distros won't care either - why should they?

Linux' 'technical excellence' in supporting every arch and every feature anyone could want (to at least some extent) is nice, but that's beside the point of which system is actually more convenient for the architectures it supports. I know if I was running a polyarchitectural (cool word) network I wouldn't use Linux no matter how much faster it was, the administrative mess of managing a plethora of different kernel sources and base packages would be a nightmare. That's how it is in Gentoo at least, maybe Debian is better after installation (but then you lose the source-based flexibility NetBSD still offers without compromise).

Re:runs on old and rare archs

[setagllib]

Can you have the same single source tree shared over NFS and have every architecture (or one cross-compilation) build for every machine on the network with any Linux distribution? Without needing any disparity in versions, and WITHOUT introducing bugs?

NetBSD's like that by design and it works. Source or binary. Hell, you can cross compile it from another operating system running on another architecture and it will still work. If there's a Linux distro out there that does it just as well (or 'better' because of that 'all the modern architectures' that 0.01% of the sysadmin population will get to look at in the next few years), well, I'd love to hear about it.

misinformation?

[diegocgteleline.es]

Linux keeps re-writing major portions of the kernel and has stability issues. It now depends on 3rd party vendors to integrate and make stable releases of the code.

Linux has always kept rewriting fundamental parts of the kernel true, and it will probably keep it that way. If not how can you explain that linux has gone from crawling in 8-ways to running in 512-cpu SGI boxes? When someone rewrites a part the kernel is for a reason, usually to do something better, and netbsd has also rewritten big parts of the kernel to get where netbsd 2.0. right now some people is rewriting fundamental parts of linux because they want to achieve realtime support. I don't see how this rewrite an be bad.

And I don't see lot of unstability issues, and I bet lot of people unsing 2.6 here will agree with me that 2.6 has been by far the stablest linux release ever. The fact that IBM has been testing linux in 32-way boxes during the whole development of the kernel has helped a lotfor that and its something BSDs can not benefit from (they don't even _boot_ on these boxes). A 32-way machine finds bugs much, much faster than a single-p4 does, it's as simple as that. That is one of the reasons 2.6 is so stable even with the new development model, people test things in those big machines before merging them in the main tree

And yes linux "depends" on distros to publish a workable system. This is how linux works, and while some people don't like it, the fact it that this way of doing things has encouraged the spread of linux,specially in the desktop - everyone can find a distro that fits to him. Do you really expect to be able to build a single base OS that 6000 millions of people will like?

Re:misinformation?

[antonakis]

Perhaps linux is somewhat unreliable because of the speed the patches are being integrated? This is not an app, this is a kernel and kernels require rigorous testing and full security audits. Take a look at the kernel changelogs.Things are moving very fast in the linux world, so fast that merges of unstable/unsecure code are frequent.Even the grsec guys were complaining at the bad quality of code which makes its way to the kernel.Nowadays, linux seems to have become a testbed for all the cool new features, while it's left for the distro makers to further stabilize and test the kernel. This job should be done before releasing the kernel.

Re:misinformation?

[Brandybuck]

Comparing kernel development to military generals is quite a stretch. The analogy falls down at every point. War is a race against time and resources, so that quick and flexible generals win. Deperate times call for desperate men. But kernel development should NOT be done by desperate men.

NetBSD stands to gain share

[idiotnot]

Why? Because it now doesn't trail in performance, and the quality seems to be better than the FreeBSD 5.x releases. (i.e. *all* of PF works, not just parts.....pf doesn't work on bridge interfaces under FreeBSD. Nor does it play very nicely with vlan support)

If you haven't tried NetBSD 2.0, you ought to. If you're looking at the now-looming death of FreeBSD 4.x and need a replacement, look at NetBSD. Also, if you have older hardware, NetBSD is probably a better choice than Linux. Glibc is very large these days, while NetBSD's libc is still pretty tight. I've been using an RC version of NetBSD 2.0 on a SS10MP machine for a few months now...zero problems, and the MP support works fine. It's also feels snappier than Solaris 9.

Where does it fit in?

[Anonymous Coward]

NetBSD is the BSD for people who don't like change, and I'm one of them. Although the system has gained many new features and has matured significantly over the past few years, the base system has largely retained the same design and layout that it has for years. Nearly every NetBSD version looks the same and behaves the same, which means you almost always know what to expect.

One of my favorite things that's come out of the NetBSD Project in the past few years is the Pkgsrc collection. Pkgsrc has been gradually evolving from a NetBSD-only 'ports' system, to a very robust cross-platform package management system. It really cuts down on a lot of work to be able to manage a handful of different Unix systems, but use the same package management scheme on each system, and keep the pkgsrc repository on a single NFS server updated with a nightly cvs cronjob.

In the BSD world, NetBSD seems to be the least driven by hype and feature creep. This makes it a real joy to use and maintain, because like I said before, you always know what to expect: a cleanly-designed, stable, functional, easy to use Unix system.

VERY handy.

[SharpFang]

We got an old SUN. 200M harddrive, poor CPU, not too much RAM, GREAT monitor, nice keyboard and mouse. A dream machine for an X terminal for our servers. But what to run on it? I tried Linux. It would barely fit. No way to fit X, a desktop manager and enough to comfortably use it. It was still possible to mount a drive over NFS and pull some binaries from there, but it was way too slow. In short, Linux sucked for it. I looked what else would work on that architecture. NetBSD? Let's give it a shot. I installed it, installed X, some basic software so it could work as a standalone workstation, not just terminal, then found enough spare diskspace, that I set up root directory and all demons necessary to run YET another SUN, a diskless workstation with equally great monitor from it (even with SWAP memory accessible over NFS, swapfile on that small drive...) So, two very nice terminals on exotic architecture, all off a 200M harddrive.

Re:VERY handy.

[SharpFang]

Oh, sure, on x86. No problem.
Find a lightweight distro for SPARC. I remember one that supported it was RedHat, I don't remember the other but it was nearly as big. "standard" Linux is helluva big, you have actively work to make it small enough to fit on such small systems. (plus don't bull me that your Slack on your 386+16M works at any reasonable speed under X. Even on 486 it's not really acceptable, even console gets slow at times.)
The thing with NetBSD was that I actually didn't have to fight for space. I was happily browsing the binary tree and kept adding components. It felt on that 200M drive, about the same as Linux on 20G one. Kernel sources, compiler tools to recompile it, NFS demons, TFTP, BOOTP, all that was needed to compile the kernel and run the diskless workstation, and if I was short on diskspace, I was just shrinking its swapfile (from original 64M to 20M or so in the end. Still left the user with some 5M for personal data...)

How is the 2.6.x kernel compilation running on your 386? :)

In a similar vein...

[ari_j]

I know a guy who built a computer into a (non-working, I can only hope, although I only saw pictures) toilet. I don't think it ran NetBSD, but it could have. Oh, and hostname = Jon (he uses Garfield character names for his network naming scheme).

netbsd is relevent cause its has a niche

[geiseri]

One awesome thing about netbsd is that it runs on some of my old hardware that never would otherwise be useful. DECstation 5000 with a sexy 21" monitor would be a paper weight without NetBSD. I can run it as an X terminal just fine. It even is nice for small stuff like my Macintosh SE/30. System 6 can only do so much, barely even run an old version of nutscrape. I have an old copy of MacX but it sucked. NetBSD, got a nice little Xterm, ssh client, etc. Fits in the server room, and bang. I have a xterm smaller than most of our VT510s ;)

NetBSD fills a need no-one else will, and because of that its relevant.

Best option for Sparc32

[Digital Pizza]

I used to use NetBSD (I forget which version it was at the time, now) on a Sparc IPX; 40Mhz, 64MB RAM, 1GB drive. I had an extra SBUS ethernet card in it and used it as my firewall, DNS, DHCP, and IMAP mail server (in concert with fetchmail).

Ran great until I started getting flooded with spam. SpamAssassin just couldn't keep up on that box; it'd still be processing the previous batch of mail when fetchmail grabbed the next batch.

I upgraded to a sparc 10 with dual 60Mhz processors, but had to move to Linux because NetBSD didn't yet support multiprocessor SPARC. It kept up OK, but 2.4 didn't support Sparc32 very well; the ext3 filesystem became corrupted with SMP enabled, so I had to go back to ext2. There seemed to be little remaining interest among the Linux kernel developers for Sparc32 anymore.

I think Solaris 10 is 64-bit only, so NetBSD may be the only option left to stay up to date on all those old Sparcs!

no brainer - commerial embedded devices

[mqx]

This is really a no brainer,

NetBSD is designed to be low footprint, highly portable, and flexible. It's the ideal BSD for embedded systems (whereas FreeBSD is suited to larger size systems and servers, and OpenBSD is unfortunately in the middle as a security oriented system, but not portable nor performance enough as NetBSD).

Licensing is a key NetBSD selling point. The problem with Linux/GNU is the GNU license which does not favour commercial embedded manufacturers who want to customise the software inside their product and (a) not have to offer the source code, and (b) not have to offer any competitive/IP/commercially-sensitive content in that source code (i.e. algorithms, device driver interfaces, etc). Despite all of the hoo-haa about the GPL, I'm afraid that companies really do like to minimise risk and lower cost by keeping their product internals as secret as possible.

Portability: NetBSD wins hands down: Linux has been ported to lots of things, but the basic architecture is not as clean. This is been shown time over again, and proven by the supported (not just "happened to be ported to") platforms of NetBSD.

NetBSD also gets to leverage the work from FreeBSD and OpenBSD, as FreeBSD really has greater commercial support in terms of device drivers and so on than either NetBSD or FreeBSD.

What NetBSD should be focusing on (in this order)

1. keeping tight BSD licenses (the kind of Theo style approach being applied to OpenBSD at the moment : to be very strict about licenses of included items) -- commercially friendly for competitive/cost reasons;

2. keeping high portability and flexibility: making sure that as new processors/platforms/drivers come along, that they can be quickly and easily supported -- commercially friendly for time to market allowing easy leverage of the existing product;

3. continually rolling in new support for hardware and security features as possible by grafting from FreeBSD and OpenBSD;

4. continually reworking and streamlining the internals to support all of the above;

5. improving the build environments (i.e. the cross compile is fantastic now), the ports system (fantastic and incredibly easy to bring third-party components in), and other things such as boot code, embedded/compressed installs, etc;

6. not getting "lost" on wasted effort for things like graphical installers, or coloured-ls's, etc;

Basically, NetBSD should continue to :

- target small/embedded devices;
- continue/improve commercial friendly;
- innovate/improve on reducing total effort to realise NetBSD onto a new hardware platform;

Re:Obviously

[bladesjester]

Do you want your toast turning into flaming death and being shot at you when your toaster blue screens?

And imagine the viruses. While it may keep me on my toes, suriken toast is not something I want to deal with in the morning...

Re:License..

[fsmunoz]

Let me begin by saying that, although I prefer the GPL to the BSD licence, both are free licences and fine by me. Actually, in a perfect world, the BSD licence should be enough (or even no licence at all...)

Now, about this "BSD licence is better for business and corporations"... it's IMHO true, but not in every way, and especially not in the way that the BSD's would gain more. From what I have saw the BSD licence is great for corporations when the idea is to *take* new code made freely available and incorporating it. But for a corporation that wants to *give* code away the GPL is, interestingly enough, better. This is so because by making it GPL the business/corporation is assured that any later improvement on the code will be available, and so it doesn't give a competitive edge to rival corporations, it more or less guarantees that from there on every implementation of the code is equal, even if being made or used by another corporation.

This makes sense; BSD licence "evangelists" are known to bring out the fact that "programmers need to eat" when dismissing the importantance of forcing the availability of the changed code. So it follows that a company will not provice ammo to rivals by allowing them to take their code and keep the changes to themselves. BSD developers are sellfishness, companies aren't.

Re:Read the WHOLE article.

[HyperChicken]

Wrong on so many counts.

NetBSD was the first open source OS to have IPv6 support. Yeah, "way out of date" IP stack.

NetBSD was the first open source OS to have USB support. Yeah, "way out of date" hardware support. Further, NetBSD allows for "Machine Independent" drivers, leading to portability far beyond other operating systems.

Not enough developers? You don't need a lot of developers. Code remains cleaner when only educated people submit features.

Unlike with other operating systems -- including FreeBSD, OpenBSD, and Linux -- NetBSD holds off on releasing features until they are stable. That's why there are few releases. This is a good thing.

It's one of the most secure operating systems in the world. Compare the NetBSD 1.6.2 security patch list to the OpenBSD 3.5 security patch list.

Re:Read the WHOLE article.

[MobyTurbo]

NetBSD isn't as feature rich as more modern kernels such as Linux, or even Free/OpenBSD. Its IP stack is way out of date;

Incorrect: NetBSD's IP stack is more scalable than all others, it broke the internet land speed record (twice).

its filesystem is way out of date

Maybe it doesn't have journaling by default (though a log file system is under development) but FFS and FFS2 in NetBSD have reliability that others could only dream of. Or you've never heard of Reiserfs and XFS loosing data?

its hardware support is out of date

That's laughable. NetBSD has some of the cleanest device driver system in existance. It was the first free OS to include USB, and it's continued porting to various platforms increase its hardware support, because thanks to its device driver infrastructure, if a piece of hardware is available for one platform, it's available to any platform that can run it.

It's not easy to get security updatesWhich is better, having to patch or use cvs a few times a year, or having to download every week new security fixes to keep from getting rooted? You decide.