Submitted by Fans of DragonFly BSD will be getting their Christmas present late this year, and plans for 1.5 have been announced. MP safe networking code, the long awaited cache coherency management system, and a port of Sun’s ZFS. Read here for more. Update: Refresh, empty cache, whatever, and check the shiny new beastie icon! And there was much rejoicing. Can we now please discuss DragonFly BSD?
anyone actually using this for day to day use? any reflections?
“anyone actually using this for day to day use? any reflections?”
I did for quite a while, but because of the problems they were having with the FreeBSD 4.x ports collection, I stopped using it for day to day things because it was a real PITA to get some of the programs I tend to use (like MPLayer for example).
A lot of work has gone into making most of the packages available in NetBSD’s pkgsrc work natively under DragonFly, and as soon as everything I want/need is available again, I plan to go back to using DragonFly for most of my day-to-day needs.
Up until just after 1.2 was released, I had used it almost exclusively and had no crashes, performace was not noticably different from FreeBSD (that is to say it didn’t seem faster or slower, and no, I didn’t benchmark anything), and except for the 3rd party software situation that arose as DF moved away from FreeBSD 4.x compatibility, it was a really solid system.
Thom, you areposting this message under a thread marked with FreeBSD new logo. While beastie, the daemon is the all BSD mascot, the new logo is just a FreeBSD thing. Dragonfly BSD has its own graphic identificator, Fred – the dragonfly. Just to clarify things.
The FreeBSD section is also the general BSD section. We can’t just go about making sections for every project on this planet, I’m sorry.
But you can have a dozen sections for different linux distributions. Ridiculous.
But you can have a dozen sections for different linux distributions. Ridiculous.
As if DragonflyBSD, in all its coolness, has as many users and developers as ANY one of those distributions.
As if DragonflyBSD, in all its coolness, has as many users and developers as ANY one of those distributions.
But undoubtedly, it’s a lot more important than any of those Linux distributions from an OS-technical point of view. DFBSD innovates OS instead of repackaging software.
The most sensical/logical view if you don’t want seperate categories would be to rename the FreeBSD section the BSD section, and use Beastie as icon.
”
But undoubtedly, it’s a lot more important than any of those Linux distributions from an OS-technical point of view. DFBSD innovates OS instead of repackaging software. ”
BSD people constantly repackage a large majority of code written for and by Linux people. Starting from GCC to GNOME written largely Red Hat developers. So…
“BSD people constantly repackage a large majority of code written for and by Linux people. Starting from GCC to GNOME written largely Red Hat developers. So…”
You really have no clue what you’re talking about do you. Please tell me the technical differences between Suse and RedHat? Other than a few configuration apps they might as well be the same, Linux is Linux. DragonFly is actually making technical innovations here. I’m not even a DFly user but at least I have some respect for what they’re doing. As for GNOME being written largely by RedHat devs, you sir are an idiot.
How about the fact that everything from package managers to init scripts vary in Linux distributions. Sure Suse, RedHat and Mandrake are pretty similar, but saying that something like Gentoo is the same thing as RedHat is pretty uninformed. To be honest even RedHat has a lot of differences as compared to Suse. Look at SELinux.
The Linux Kernel is the Linux Kernel across the majority of distros.
Yup, but the kernel aren’t the same kernel across the distros. They are all linux kernels but not the same kernel.
But yes, I do think there are too many linux distro categories here on OSNews.
Whether or not it is fair must depend on number of news and users for each category. I think there are too many categories.
As a way of making people happy OSNews might want to change the icon for the BSD section. It just might work out better. At least there is a hope.
But all the kernel patches can be installed on all the kernel of all the dists, or atleast almost. To compare different init-scripts, package formats and installers with a “whole new os” (or atleast the system part of it) if stupid. I have no idea how many dists there is which are based on debian, but to say they would be more important than a new os?
But it’s “just” another *BSD I can hear people claim.
I don’t really care, as long as the news are being brought. That’s more important than the categories, right?
Relax people, it’s just an icon
I agree. Too many posts about an icon, and not enough about the article.
“”You really have no clue what you’re talking about do you. Please tell me the technical differences between Suse and RedHat? Other than a few configuration apps they might as well be the same, Linux is Linux. ”
I am afraid you are the one who needs to brush up your knowledge. Linux as the BSD people are very fond of pointing out is only the kernel. Starting with everything from SELinux support to cluster filesystems like GFS and Red Hat directory server, Fedora or RHEL is entirely different from the SUSE products.
BSD people constantly repackage a large majority
of code written for and by Linux people. Starting
from GCC to GNOME written largely Red Hat
developers. So…
Just because something is GPL’ed doesn’t mean that it is Linux. GCC existed years before Linux let alone RedHat. Now technically RedHat does a lot of work on GCC becuase they bought Cygnus Support but it is the FSF that is responsible for releases of GCC not RedHat or any of its minions.
I ran and supported GCC on BSD UNIX years before Linux was on the scene.
“Now technically RedHat does a lot of work on GCC becuase they bought Cygnus Support but it is the FSF that is responsible for releases of GCC not RedHat or any of its minions.
”
No sir. It wasnt. Red Hat Micheal Tiemann OSI president wrote the first GCC C++ Compiler and ported it to Windows as part of cygnus operation which incidentally is well known as the first free software company. Red Hat people are steering commitee members and during the egcs fork merge effort, FSF explicitly agreed not to interfere in any of the decisions with respect to GCC in return for it being a GNU project
Most of the code is not, as far as i know, written specifically for linux, in fact, i would propose the thinking that most linux distributions repackage a large majority of code written for GNU (gcc) and free software purposes (ie X-Windows)
Agreed!!!!
I thought about saying something similar; however, I didn’t because.
1) My statement wouldn’t have been as short or as sweet.
2) Didn’t feel like playing with the OSNews staff.
“But undoubtedly, it’s a lot more important than any of those Linux distributions from an OS-technical point of view. DFBSD innovates OS instead of repackaging software. ”
Indeed. I’m not a DragonFly BSD user, but I find it far more interesting from a technical point of view than Yet Another Linux distro. Perhaps it’s time to revamp the categories of OSNews and clean out a few the least used and instead add more relevant ones?
Perhaps it’s time to revamp the categories of OSNews and clean out a few the least used and instead add more relevant ones?
We do that continiously. Not too long ago we created RISC OS and Zeta categories. DragonFly simply doesn’t get enough news and attention to get it’s own databse category, I’m sorry. Next thing you’ll know we need to create seperate categories for AmigaOS 1, AmigaOS 2, AmigaOS 3 and AmigaOS 4.
Thom, now you’re just being plain silly. Yuo can’t possbily compare 3 dead and 1 zombie version of an OS to 4+ BSDs that are all alvie and kicking. But, as you say, there’s not enough news for DragonFly BSD for the time being.
“But you can have a dozen sections for different linux distributions. Ridiculous.”
Heh. I’m a big fan of DF, but I’d be suprised if there is even one thousand people using it. Too young, too small. For the time being, I think it fits quite happilly where it is, in the “FreeBSD and other BSDs” section.
The FreeBSD section is also the general BSD section. We can’t just go about making sections for every project on this planet, I’m sorry.
Don’t be sorry Holwerda. Just make a BSD section with the BSD logo and let all BSDs be a part of it. Its logical.
Maybe someone can clarify for me, but the OpenSolaris CDDL doesn’t seem to be compatible with the revised BSD license under which DragonFly is released. Maybe I’m missing something?
http://www.opensolaris.org/os/licensing/opensolaris_license/
Still, I’d really like to see a new FS option for a BSD. I was kind of hoping for HFS+, myself, but ZFS seems neat, too. UFS/FFS is old, slow, and very difficult to deal with on large filesystems (fscking a large partition takes a loooong time….I’ve seen 140GB take nearly two hours). UFS2+softupdates still isn’t data-safe (since it essentially requires async operation), and doesn’t improve performance. 1TB+ filesystems. woohoo. How long have other filesystems supported that?
I think, it might be same as how all BSD are doing with GPL stuff, nve (nic driver, NVIDIA license) and other stuff in the system. I might be wrong as I don’t know about CDDL that much, thought.
You complain UFS is old and slow, but were hoping for HFS+? We’re talking about Apple’s HFS+, are we not? The biggest relic of a filesystem still in common use after FAT32?
HFS+ was completely new for OS8.1. It’s a very nice modern filesystem.
HFS wouldn’t help matters in BSD, no. 🙂
To put it another way, there’s good reason why Apple did work to make HFS+ Unix-friendly, rather than try and improve UFS for use with OSX. NeXT/Openstep used standard FFS.
HFS+ is simply an extension of the old HFS code. It’s the same basic filesystem that’s been in Macs forever, it’s just that the block address sizes have been extended to cover larger filesystems. In a very real way, HFS+ is to HFS as FAT32 is to FAT16.
The first Mac filesystem was MFS.
It’d be more accurate to say that HFS+ is to HFS as NTFS is to HPFS (which is 16-bit, and 100% MS….why IBM went to JFS in later versions of OS/2). NTFS is 32-bit, includes journalling, is currently being developed and improved, etc. etc.
Still, having dealt with macs running HFS+ since 8.1 was brand new (afaik, one of those is still in use at my former job, doing ProTools work every day), I can say that it is a) fast, and b) very stable.
The first Mac filesystem was MFS.
I never said HFS was the first. I said it has been around forever. Wikipedia shows that HFS was introduced only a year after MFS (in 1985), so my point still stands.
NTFS is 32-bit, includes journalling, is currently being developed and improved, etc. etc.
IIRC, NTFS is based on HPFS, but is significantly different in many regards. NTFS is a a generalization of the HPFS design. HFS+, in contrast, keeps almost precisely the same structure as HFS, except it makes various control structures larger. Beyond that, HPFS is a far less primitive base for a modern filesystem design than HFS!
I can say that it is a) fast, and b) very stable.
I wouldn’t call anything filesystem related on OS X “fast”. It’s hard to give numbers relative to other FSs on OS X, because its UFS implementation is poor, but compared to Linux on comparablely fast disks, OS X is much slower for things like compiles. Compiles exercise the filesystem much more heavily than media processing, since they touch a lot of metadata and a lot of small files, as opposed to just touching the extent maps in a few large files.
Speed aside, HFS+ is certainly a less interesting filesystem from an FS theory point of view than either ZFS or Reiser4.
To put it another way, there’s good reason why Apple did work to make HFS+ Unix-friendly, rather than try and improve UFS for use with OSX. NeXT/Openstep used standard FFS.
No, its more the need to for file forks support and the need for case preserving but case insensitive support – that and the fact that HFS+ isn’t that broken, as much as the nay sayers would love to make out.
Sure, its the not the prettiest file system out there, but at the same time, is it worth replacing it with something else? I mean, if they’re going for something that has ‘teh cool’ factor, they may wish to talk to SUN and see if they can port ZFS over to Mac OS X.
As for DragonFly, its looking like its really ontrack, and the gamble to try a new route to MP capabilities has really paid off in the end rather than the approach which the FreeBSD team took – thats not to say that Dragonfly got everything right, but at the same time, when you compare the resources that both teams have got, they’ve come a long way, very fast.
Maybe someone can clarify for me, but the OpenSolaris CDDL doesn’t seem to be compatible with the revised BSD license under which DragonFly is released. Maybe I’m missing something?
You are.
1) CDDL is a file based license, not everything under DragonFly is under a “different” BSD license.
2) The DragonFly license is file based too.
3) Matt is no stranger to licensing I am fully confident that he knows what he’s doing.
4) It is my personal belief that even the modified DragonFly BSD license could be considered compatible since again CDDL is a file based license…
No further comments
I second that! (in regards to the subject)
Speaking of Logos, is everyone using the new FreeBSD logo except the FreeBSD project page itself? I liked the old style FreeBSD page better, the new one tries too hard and is way too ” in your face “.
Still, I’d really like to see a new FS option for a BSD. I was kind of hoping for HFS+, myself, but ZFS seems neat, too. UFS/FFS is old, slow, and very difficult to deal with on large filesystems (fscking a large partition takes a loooong time….I’ve seen 140GB take nearly two hours). UFS2+softupdates still isn’t data-safe (since it essentially requires async operation), and doesn’t improve performance. 1TB+ filesystems. woohoo. How long have other filesystems supported that?
http://www.freebsd.org/marketing/os-comparison.html
[quote]
Linux is well known for its reliability. Servers often stay up for years. However, disk I/O is non-synchronous by default, which is less reliable for transaction based operations, and can produce a corrupted filesystem after a system crash or power failure.
[/quote]
http://www.freebsd.org/marketing/os-comparison.html ?
lol, that one is so old it’s not even fun, I do like FreeBSD but currently I guess linux 2.6.x wins most or everything when it comes to speed.
How hard is it to change a flag in /etc/fstab for your filesystem if you want whatever behavior?
”
http://www.freebsd.org/marketing/os-comparison.html
[quote]
Linux is well known for its reliability. Servers often stay up for years. However, disk I/O is non-synchronous by default, which is less reliable for transaction based operations, and can produce a corrupted filesystem after a system crash or power failure.
[/quote]”
That link is choke full of misinformation. Just a few examples,
“Linux does not use any version control system so all bug-fixes and enhancements must be emailed back and forth on mailing lists and ultimately submitted to the one person (Linus) who has authority to commit the code to the tree.”
Wrong. Linux uses a distributed source code management system called git and for several years before that bitkeeper. You can find all the change sets at http://kernel.org/git
“[Bad]The Linux ext2 filesystem gets its performance from having an asynchronous mount.”
Incorrect. Linux ext3 filesystems mount it the same as freebsd mount by default
“The situation has improved somewhat recently and the 2.4 release of the Linux kernel introduced a new virtual memory system based on the same concepts as the FreeBSD VM system.”
2.4 is not recent by any stretch of imagination
“This problem is compounded by the fact that distributions like Red Hat tend to turn on notoriously insecure services by default.
”
Outright lies. Majority of daemons on RHEL or Fedora only listens to the local loopback device and not to any external network by default. They both are the first mainstream operating systems in the world to come with MAC based security by default using SELinux and hence are much more secure than any version of BSD
Lol. That comparison wasn’t even accurate when it was first published about five or six years ago, and now it is woefully out of date.
Tom, what people say is that the new FreeBSD logo is FreeBSD _only_. The common logo to cover _all_ BSD’s is the deamon. You’ve been asked to only use the FreeBSD logo on FreeBSD articles and not on *BSD articles. Is that so hard to understand?
Yuo can’t possbily compare 3 dead and 1 zombie version of an OS to 4+ BSDs that are all alvie and kicking
Dead? Zeta has sold more copies of its OS than Be Inc has done. There are more regular Zeta users in the EU alone than DragonFlyBSD has over the world. For RISC OS, there are more users of that in the UK alone than DragonFlyBSD has over the world. As for Amiga– the Amiga is something with such an heritage, I wouldn’t be too dismissive of it if I were you.
Tom, what people say is that the new FreeBSD logo is FreeBSD _only_. The common logo to cover _all_ BSD’s is the deamon. You’ve been asked to only use the FreeBSD logo on FreeBSD articles and not on *BSD articles. Is that so hard to understand?
Take a look at the database category’s name. It’s “FreeBSD and other BSDs”– but FreeBSD constitutes the major part of the category. THAT is why it has the FreeBSD logo.
I don’t hear people complain over AROS being placed in the Amiga category, even though AROS and amigaOS differ more than DragonFlyBSD and FreeBSD. I find this discussion quite pointless and really, as if there aren’t more important things in the world.
But fine, then you guys get the generic OS icon. A fork of FreeBSD 4.x gets the generic OS icon even though there’s a general BSD category. Yes, that makes a lot of sense.
I love the invented user statistics and the “sold more than BeOS” argument.
What people have been asking all along, and apparently you have missed, is that you don’t use the FreeBSD logo for all BSD projects, but rather a BSD icon for all of them.
What people have been asking all along, and apparently you have missed, is that you don’t use the FreeBSD logo for all BSD projects, but rather a BSD icon for all of them.
Apparantly you don’t understand how it works in the world of OS communities. When we revert back to the old FreeBSD/beastie logo for the “FreeBSD and other BSDs” category, I will get emails from FreeBSD people who demand I put up the new logo– exactly what happened not too long ago when the new logo was announced.
And I’m sorry, but FreeBSD items make up for 95% of the content in the “FreeBSD and other BSDs” category, so they get to “choose” the icon.
Now, I’ll place any news related to DragonFly BSD in the general category– which is much less appealing and defining, and it degrades DragonFly BSD to the hobby status, but I have little choice.
Now you’re doing that just to spite us. This is the least sensical option.
Now you’re doing that just to spite us. This is the least sensical option.
No, as I explained, you guys leave me with no other option. You don’t want it under the “FreeBSD and other BSDs”, because for reasons I still do not understand, it doesn’t fit that description. Fine, I’m okay with that.
However, then you only leave me with one other option, and that is the general category, because making a new category specifically for DragonFly BSD is out of the question, and changing the icon for the “FreeBSD and other BSDs” category to a general BSD icon is also out of the question, because 95% of the stories in that category ARE FreeBSD stories.
May I suggest a compromise? Add a topic specifically for Other BSDs (DragonFly, DesktopBSD; PcBSD etc etc)? That way the FreeBSD topic would be pure FreeBSD which makes sense given the fact that FreeBSD is the dominant BSD. If anything, the “Other BSDs” should end up with the smallest of the traditional BSDs (either NetBSD or OpenBSD).
By the way, you still call the original BeOS category “Zeta, BeOS & Derivatives” and since you’ve created a specific Zeta topic you may wish to change that.
“Yuo can’t possbily compare 3 dead and 1 zombie version of an OS to 4+ BSDs that are all alvie and kicking”
“Dead? Zeta has sold more copies of its OS than Be Inc has done. There are more regular Zeta users in the EU alone than DragonFlyBSD has over the world. For RISC OS, there are more users of that in the UK alone than DragonFlyBSD has over the world. As for Amiga– the Amiga is something with such an heritage, I wouldn’t be too dismissive of it if I were you.”
I was referring to AmigaOS of which I have over 10 years of experience of. AmigaOS has indeed quite some heritage, but I failed to see how you could compare giving DragonFly BSD its own topic to giving AmigaOS 1.x its own topic. That is ALL. There is no need fo you to be so defensive especially since you’ve misinterpreted me. Perhaps I was not clear enough.
The goal of single system image is what makes DragonFly BSD interesting. An industrial-strength, truly clustered, free UNIX. This is what will set it apart from Linux and the other BSDs.
But ladies and gentlemen of this supposed OSNews site, I have one final thing I want you to consider: Ladies and gentlemen, http://content.answers.com/main/content/wp/en/1/1d/Chewbacca.jpg this is Chewbacca. Chewbacca is a Wookiee from the planet Kashyyyk, but Chewbacca lives on the planet Endor. Now, think about that. That does not make sense! Why would a Wookiee — an eight foot tall Wookiee — want to live on Endor with a bunch of two foot tall Ewoks? That does not make sense!
But more important, you have to ask yourself, what does this have to do with this case? Nothing. Ladies and gentlemen, it has nothing to do with this case! It does not make sense!
Look at me, I’m a geek defending the general *BSD mascot icon, and I’m talkin’ about Chewbacca. Does that make sense? Ladies and gentlemen, I am not making any sense. None of this makes sense!
And so you have to remember, when you’re in your computer rooms deliberating and conjugating the Emancipation Proclamation… does it make sense? No! Ladies and gentlemen of this supposed OSNews site, it does not make sense.
If Chewbacca lived on Endor, you must use the general *BSD mascot icon! The defense rests.
In regard to the debate about the FreeBSD icon, I must say that Eugenia was very opinionated as editor of this site, and whenever her opinion solidified on a matter, she would not budge, but when she expressed her opinion, she sounded rational. Her arguments made sense, even if you disagreed with them. I must say I think putting all the BSDs in a single, general BSD category makes the most sense. Thom, I’m sorry but you haven’t articulated your to the satisfaction of your readers. Maybe we all are in the dark, but frankly, I think you are being immature and childish.
You have two ways you can run the site: you can listen to the feedback the members of OSNews give you, or you can dictate the policy based on any irrational whim. You could justify the current icon system to your readers’ intellectual satisfaction by saying “because I said so,” but please don’t pretend that your argument makes sense. If the form of the website is dictated by self-centric feelings, rather than thoughtful rational design, you run the risk of alienating your readers,
because a tech news site is a poor example of a space where feelings are paramount. If you want a personal web space, I can suggest one to you: http://www.blogspot.com/
Though I can’t offer any sort of objective “betterness” measure, the consensus here stands STRONGLY against your argument, and strongly FOR the creation of a general BSD category. They are all derived from a common code base, and continue to carry the BSD moniker. Additionally, and probably most importantly, they all occupy a certain mindspace for the readers of your site. I can scant think of FreeBSD without association to the other BSDs.
Claiming that the FreeBSD folks would balk at this idea presumes their immaturity. Frankly, I don’t believe you they would have any problem whatsoever with this. If the developers of FreeBSD were all clones of Theo DeRaadt, then we might have a situation, but there’s no reason to postulate about negative feedback which is only marginally likely, when the readers of your site have given you real honest-to-goodness feedback.
Upon a second reading of your posts, I realize the problem with the FreeBSD logo. The problem is really that no logo has been formalized as an offical common BSD logo. De facto, though, they all have a demon, or similar as their mascots. http://www.freebsd.org/ STILL has the beastie on the front site, so it gives credibility to the notion that we can easily designate the BSDs as a group with a demon.
This is all intended as constructive criticism, Thom. I thank you for all of the hard work you and others have put into OSNews. The new moderation system is great! I hope to see you grow as an editor, and I hope that this disagreement between you and the members of OSNews finds a resolution that is satisfactory to all, even if it’s not exactly what everyone wanted. (Your more mature readers will understand the necessity of compromise.)
“This is all intended as constructive criticism, Thom. I thank you for all of the hard work you and others have put into OSNews. The new moderation system is great! I hope to see you grow as an editor, and I hope that this disagreement between you and the members of OSNews finds a resolution that is satisfactory to all, even if it’s not exactly what everyone wanted. (Your more mature readers will understand the necessity of compromise.)”
And constructive criticism it is. I’d like to emphaszie that I appreciate the hard work the OSNews folks put into the site and that it is the first stop for me during my news sweeps. Thom, I understand that it must be difficult to fill ou Eugenia’s shoes, but never ever forget that your greatest resource is your readers. Learn to listen to them. Even if they post anonymously. 🙂
Stop being a childish little prat Thom! The little shiny ball with two horns sticking out doesn’t real mean “BSD” to a lot of people. Just because FreeBSD Released a new icon, and “95%” of the articles are FreeBSD related under the BSD category doesn’t mean that this is appropriate. As mentioned earlier this isn’t even the officeal FreeBSD Logo! Please just change it to an icon that reads “BSD” or the old “beastie”. Please!! You are just alienating BSD users to further your own ego.
You guys got what you wanted– we probably created a whole new set of enemies because we eliminated the Darwin category/logo.
Stop being a childish little prat Thom!
Excuse me? Childish? Who started this silly discussion over an *icon*?
and “95%” of the articles are FreeBSD related under the BSD category do
For what it’s worth, and I’ve only mentioned it like, what, 12874239340438682356348583045 times, there was no such thing as a “BSD” category. We had a “FreeBSD and other BSDs” category. But now, you have your “Other BSDs” category.
And there was much rejoicing.
“Stop being a childish little prat Thom!…Please!! You are just alienating BSD users to further your own ego.”
Too bad I’ve run out of mod points. I’ll just have to shortlist this one so I can MOD IT DOWN!
This entire article was hijacked by the Free BSD logo debate. Doesn’t anyone find it interesting that they are going to rewrite huge amounts of their core OS in version 1.5 so that it will work over clustered systems and tie in ZFS into this same work?
This is really quite a bit of important work! Being that this is an OS News site, I would think the level of work that is being done in this project would be exciting and certainly be noteworthy! More noteworthy than the type of icon the article used.
If the DragonFly BSD project is successful in geting their goals completed for the 1.6 release – including getting the cache coherency scheme functional (clustered systems where processes can be migrated based on load) and succeed in getting their code fully MP safe using LWKT, and (perhaps) XIO all whilst using ZFS for their filesystem – would their then be considerable interest in this operating system?
In other words, if DragonFly is successful in becoming a technically superior operating system, is there a possibility that this operating system would make significant inroads. Or is it possible that no matter how technically correct or advanced an operating system is, if it doesn’t have enough hype it won’t flourish?
Thank you! I wish I could mod you up.
In what ways would it be “technically superior” to other operating systems?
Clustering, clustered filesystems, etc have been around
for quite a long time and operating systems like OpenVMS
and Linux are quite good at them.
SSI shared memory clusters are possible today with Linux
using a page based coherency protocol. I don’t think
they’re particularly practical because non-transparent
cluster software is really mature now and one actually
needs to program for a cluster in a non-transparent
way (far more important than even a slow NUMA) to get
decent performance.
However I’m not sure about what sort of SMP scalability
DFBSD would have if all code was fully MP safe using
LWKT as you say, nor advantages of XIO or ZFS.
Now the iconbusiness is out of the way, let’s discuss.
ZFS no advantages? ZFS has numerous advantages over older filesystems, all to do with data-coherency, capacity and speed. Being one of the first BSDs to implement it would be a good thing.
And if the LWKT model works out, it would provide an excellent basis for SSI, wether or not Linux does it as well.
ZFS is not an *advantage* as such because it is available
on another system.
Sure it may be an advantage for DFBSD + ZFS versus Linux + something worse, but if you’re talking about innovation
and advantages of DFBSD project then ZFS is not so interesting.
Now if you can do something with DFBSD + ZFS that you
cannot do with Solaris, that too would be interesting.
I fail to really see how LWKT (lightweight kernel
threads?) is a basis for SSI…
Well that is the the reason (cache locality) that Light Weight Kernel Threads were developed. The processes themselves are per CPU and are migrated (and communicate via a light weight messaging protocol.) This suits itself to clustering because you have an increased L2 utilization and a means of migrating processes to systems with heavy loads to other non-taxed systems in a cluster. Cache coherency systems are not new, however they are new to this system and because of the way the system is developed it has a potential to perform and scale really well in a clustered environment, talk on ZFS which Matt Dillon thinks will work very well in such an environment and you may end with something that works VERY well.
Per-CPU scheduler and CPU affine processes which migrate
between CPUs are nothing new. Nor would it particularly
be of help for a cluster because migrating processes
between CPUs with a per-cpu scheduler is by no means
equivalent to migrating processes between nodes in a
cluster.
Now if the cluster is SSI on a page level (like Linux’s
virtual iron capability AFAIK), then you may not have to
explicitly migrate processes but they could migrate
naturally according to the scheduler. This also seems
like a pretty bad idea unless the scheduler is very
specialised, because you could have a process with local
memory scattered all over your cluster.
“and because of the way the system is developed it has a potential to perform and scale really well in a clustered environment”
Well yes, we are talking about software cache coherency
schemes for SSI clusters. So what exactly is it about
the way this system is developed that gives it this
potential?
“talk on ZFS which Matt Dillon thinks will work very well in such an environment and you may end with something that works VERY well.”
I recall him saying ZFS is a clustered filesystem or can be used over a cluster. This is not the case.
Incorrect. Linux ext3 filesystems mount it the same as freebsd mount by default
Anyway a Linux filesystem is still currupted after power failure, I am talking about Fedora now. I’ve never seen FreeBSD crashing on that.
They both are the first mainstream operating systems in the world to come with MAC based security by default using SELinux and hence are much more secure than any version of BSD
Yeah right. Never heard of TrustedBSD i guess.
I think the key thing you’re missing here is the “by default”.
Actually, the progress made by the TrustedBSD project seems to have largely been merged back into the mainline FreeBSD system starting with 5.0, possibly earlier.
Since 5.3 was considered production quality, when was it shipped? November 2004. When was RHEL4 shipped? February 2005.
From that data it seems like FreeBSD was the first mainstream system with MAC-based security.