Username or EmailPassword
Ugh. The site is so slow ATM. Must really be getting Hammered
(Heh, it was that or a Nathan Fillion joke)
End of this year I'm looking at putting together a server for my lounge room, its tempting to get an ATOM based CPU/Motherboard combo and use dragonflybsd. The hammer filesystem will be a great file system for the several terabytes I'm looking at setting up to store movies and so forth on.
For those interested, here is an interesting white paper on the matter:
I'd love to see it being ported to MacOS X; the problem with ZFS is that it requires a significant amount of memory due to the nature of its design, so I wonder if HAMMER is the better way to ago around it given that Matthew (in the PDF) states that the difference between Hammer and UFS is small in terms of performance. Edited 2008-07-24 04:21 UTC
If you have a look at the following:
It doesn't appear as though HAMMER is designed for desktop systems; it reserves quite a lot of space that it uses for a buffer and you won't get free space back immediately after delete operations - you'll need to wait for the cron jobs to run (that you must set up yourself) before the file system is cleaned up.
It certainly looks like something you would deploy on clustered enterprise hardware but I'm not sure it would replace ZFS on the Mac/FreeBSD desktops any time soon, though I do share your ZFS memory requirements concerns.
The OP wants to use it on a multi-TB fileserver for storing movies and music. Pretty much an ideal use for HAMMER. It's not like you'll be deleting your library all that often.
FreeBSD 7 and DragonFlyBSD 2 on the same year ... SCHED_ULE and HammerFS ..
I just hope hammer gets ported to FreeBSD ASAP !
And boy .. am i looking foward to the LiveDVDGSoC or what!
Hope they can work it out before the end of the year
Thanks Matt and the DF crew and _Congratulations_ on your the 2.0 release !!
This was my first BSD -- installed it, and the bootstrap loader consistently crashed. Talked to the people on #dragonflybsd, and they looked through the sourcecode, and it's either a hardware error or a programming error. They asked me to install FreeBSD just to see if the same problem occurs. Now happily running it, got wifi internet natively, X running easily.
If my memory serves me, during the development of the FreeBSD 5-series, there were many architectural decisions that the lead developer of DFBSD disagreed with, and he decided to start his own fork based on the 4-series code. Since the advent of the 6-series, and now the 7-series, FreeBSD has undergone many, many corrections to the (perceived) inferiorities which plagued the 5-series. I know I'm painting with broad strokes, but I do remember specifically the reasons the DFBSD project was created. What is the reason nowadays, assuming the perceived inferiorities have been acknowledged and corrected, for the continued existence of the DFBSD project? Having seen the benchmarks, I'd say that there's a great difference between the 5-series and the 7-series, especially in the areas which originally prompted the fork, namely SMP scalability, and the threading model.
Do the two projects benefit from a synergistic benefit that the organizational overhead of incorporating all the developers into one project would otherwise cancel out? Is DFBSD intended as a testbed for experimental OS technologies, without the pressure of having to keep your code in sync with a mainline branch? Is development carried out in order to compete with, and spur innovation in the two products, or is DFBSD developed for the sheer joy of having your own OS? (A very reasonable and I'm sure unique joy that's not available to lesser mortals such as myself.)
I like to see variety in OSS projects as much as anyone else, but given that there are finite resources behind the development of OSS, I'm genuinely curious whether the benefit is intended to be public or private, or a mix of both.
(NOTE: If you read this as me telling anyone else how to spend their time or efforts, you will surely and rightly think I'm a presumptuous asshole, but it's in no way my intent to tell others what to do. I'm just curious, that's all. Any answer, even personal ones, or "no particular reason," would satisfy me.)
If anyone knows, please let me know!
Oh, and congratulations to the DFBSD team for their 2.0 release! Edited 2008-07-26 04:06 UTC
In reply to: Question about DFBSD's charter ..
(yeah .. I pushed the wrong button )
I think you might find most of the answers to your questions in here:
Based on those goals .. I'd dare to say .. those are reason for the continued existence of the DFBSD project
HAMMER as serious replacement for UFS seems (at least for me) is reason enough to keep the "separated" efforts ...
Is development carried out in order to compete with, and spur innovation in the two products, or is DFBSD developed for the sheer joy of having your own OS?
Again .. DF has different goals .. HammerFS excplicits innovation ..
Now ... "compete"? .. I hardly see anything done under in the BSD world (and ultimately .. under the BSD license) in "compete" terms ... every other BSD benefits from whatever improvements goes on in the other BSD's ...
I see it more like "each one has it's own way to push CS one step forward ..." and I really fail to see how that can be a bad thing.
Just my point of view.
Regards Edited 2008-07-26 23:50 UTC