James Chacon of the NetBSD release engineering team has sent a report covering the status of the NetBSD 2.0 branch to the netbsd-announce mailinglist. The report contains a schedule for the release cycle, and a list of 2.0-specific bugs that need to be closed. This is still a good time to help us making this the best NetBSD release ever, by trying out the latest snapshots, and reporting bugs.
NetBSD 2.0 status report
Submitted by Daniel de Kok 2004-07-28 NetBSD 16 Comments
Did I get that right?
NetBSD 2.0 will be released in August? Hooray!
I know that NetBSD is a great research and development platform, especially due to its extreme portability, but sometimes I get a feeling that it’s become quite a small niche project when compared to both FreeBSD or OpenBSD. Or am I totally wrong?
If you want security, choose OpenBSD. If you want performance, or a BSD desktop with decent features (multimedia, 3d), usability and enough available software, choose FreeBSD (or maybe in the future DragonFly BSD?).
By the way: weren’t the NetBSD people also going to announce their brand new logo before the end of July?
Probably not this one though: http://mail-index.netbsd.org/netbsd-advocacy/2004/04/01/0000.html
I always thought it was, from day one. Not that that is a bad thing tho. NetBSD is being used in alot of embedded products, check out the company Wasabi Systems – http://www.wasabisystems.com/ – I believe they were founded by NetBSD developers. OpenBSD may claim the title of “most secure” O.S. but NetBSD is hardly all that shabby either. Plus its faster and more scalable then OpenBSD. To some people, especially with older hardware more sensitive to performance, that matters. Personally I’m very fond of OpenBSD, mostly for feel more then anything else, but NetBSD is quite sweet too.
NetBSD makes just as snazzy a desktop as FreeBSD does. 🙂 Honest. Theres simply less handholding is all.
However, how scalable is NetBSD when compared to the current FreeBSD in development (5.x)? the documentation seems rather thin; I understand NetBSD has adopted the scheduler activations along with the nifty new threading, however, there isn’t much about its SMP capabilities.
NetBSD makes just as snazzy a desktop as FreeBSD does.
I suppose so. But it’s the lack of interest in desktop software and the lack of such features as, say, 3d support that restrict people’s potential interest in using either NetBSD or OpenBSD more as a desktop system. Probably the NetBSD/OpenBSD developers are not even very interested in something like 3d videocard support? Their focus tends to be elsewhere than serving the needs of typical desktop users.
I believe theres a wraper of sorts that allows NetBSD to use the FreeBSD NVidia driver. http://kerneltrap.org/node/view/671 for more info. A quick googling showed that I was correct.
NetBSD 2.0 will indeed have scheduler activations (M:N threading model) kqueues and other goodies. It also supports SMP for various archs . However I believe that the giant lock is not out of the kernel yet. This definitely impacts scalability. This IMHO involves a lot of work and also careful designing beforehand. NetBSD may not be the fastest to get a feature, but when it does something, it does it right. Given the smaller developer base it may take longer than others to actually get “fine grained” smp, but when it does you can rest assured that it will be of very high quality.
I test netbsd 2.0 wherever I can, and pkgsrc provides a very robust source harness for third party applications. It functions great as a server or as a desktop, and I’d recommend interested parties to certainly give the 2.0 release or the beta a try.
summary from above report: “NetBSD now scales better than even FreeBSD 5!”
The often repeated phrase is: FreeBSD for performance, OpenBSD for security, NetBSD for portability. After hearing this, people may think “OK, I’ll run FreeBSD as my desktop and server and OpenBSD as my firewall, but I don’t have computers with various cpu architectures so I don’t need NetBSD.” What these people don’t realize is that porting code to other platforms exposes bugs and bad design decisions that would otherwise be missed. Fixing such issues later can cost lots of effort that could have been saved had the problems been detected and sorted out earlier.
Because NetBSD has much less resources than the more popular FreeBSD, good design (both in overall planning and in small details) is very important. Portability makes it necessary to drop nonessential features and to make the implemented features as flexible and functional as possible. Portability is also a pragmatic test bed where design decisions and code quality are constantly tried out. So, as a result of portability, NetBSD has become small but flexible and powerful in it’s features. It is fast and it performs very well under heavy load. The quality of code makes NetBSD also stable and secure.
The downside of portability (and of low development resources) is that the release cycle of NetBSD is way too long. But when a new major release of NetBSD comes out, it’s something worth waiting for. I’d say that if FreeBSD is for performance and OpenBSD is for security, then NetBSD is for both performance and security (plus it’s portable).
Is it because a real comparison with people who know how to tweak things right would show completely different results than this Fefe thing that people don’t realize how crappy that comparison is?
What that scalability issue is truly about is not comparing operating systems but to figure whether the user is intelligent enough to question things in it’s surrounding environment. So if you use that as a reference and can’t question it, I guess we all know where that places you “tech_user”..
i beg to differ. any intelligent user will question such things are benchmarks and reports on operating characteristics. i happen to belive that the author of that particular report was quite thorough to *reasonably* coax as much out of each OS under test, using tests which are not too different from common tasks to which these operating systems are put. as such i think it is a valuable work.
*reasonable* – most admins/users/tech_users don’t have the resources to “tune and tweak” their favourite OS to perform better than published benchmarks. sure you could tweak the NT kernel to give you better memory allocation or thread switching than OS XYZ … but how many people will do that unless the design of the OS makes that easy. that is why the published docs are useful – the author only goes to reasonable lengths to tune the OSes. and its all on his site… the methodology, any recompilations, and specific options (such as static vs dynamic binaries) ..
but i agree with you – most published benchamrks reflect circumstanes which rarely occur in common use and as such are not useful for those wanting to perform those tasks.
i myself am using teh netbsd-current and it performs admirably.
there are other reasons to use netbsd over the other bsds… and “by design” is one of the better reasons.
Anyone running NetBSD on a G3 iBook/PowerMac or iMac? whats its performance like compared to MacOS 10.3.4 running on the same machine? I’m running MacOS 10.3.4 on an iBook G3 366Mhz and everything is ok in the performance department, however, I’ve always wondered whether the performance would be better under NetBSD or OpenBSD.
Anyone with any experience with those machines?
I have been tinkering with NetBSD for a few years now and pefer it over any Linux distro. I currently run the 1.6.1 DEC Alpha port on my webserver and of course it performs flawlessly. 1.6 was a significant improvement over 1.5 so my hopes are 2.0 will be even better. Its a very powerful OS and fits easily on a 3″ CDR.
After trying out gentoo, slackware, debian and freebsd, I tried installing NetBSD on my old machine. It was by far the fastest and most stable. It was also more memory efficient. For low-end machines NetBSD accompanied by XFCE is a good choice for a fast, user friendly Internet desktop.
I’ve found NetBSD to be a great system on older hardware, especially a couple of Pentium laptops I had lying around. I initially tried using Debian, but once I noticed that ‘apt-get’ ends up paging horribly on systems with <64MB of RAM, I started looking for alternatives. NetBSD proved to be a perfect match: very memory and drive-space efficient, good performance, and a default install that brings in effectively no cruft.
Since then, I’ve used it in some embedded and multimedia projects, and never regretted the choice. Plus, I’ve been able to use it to “upgrade” a couple of old Macs my family had lying around, and am still on the lookout for an old SPARCstation or BeBox to load up with it.