FreeBSD FreeNAS drops FreeBSD in favor of Debian

[vermaden]

Originally Posted by votdev

Hello,
seems i have to say something about the current situation.

Since long time many users want to see more and more features in FreeNAS. But this is not possible as it is designed at the moment. The intention of FreeNAS was to be smart and simple. Because the roots of FreeNAS is m0n0 the complete design is to be compact. But for a NAS and all it's features we come to a situation that the complete core of FreeNAs is not able to meet that requirements anymore. It is really a pain to add new features and i'm not willing anymore to deny every request. It is hard to tell someone that it is not possible to add these or those features because of technical reasons if you do not have an overview of how FreeNAS works internally.

Some months ago i started to upgrade FreeNAS to FBSD8.0. I also thought about how to remove all these limitations of the previous versions to make it more enhanceable. The more i thought about it the more i came to the conclusion that it will be necessary to reimplement everything to get rid of these limitations.
The first is there will be no difference between 'embedded' and 'full' anymore. This is one of the reason why FreeNAS is so complex. Every script needs to ensure that the requirements are fullfilled to run without problems. Also the complete build scripts need to reimplemented because i do not want to go the same way as in the current versions. Right at the moment only the required files (binaries and libs) are copied. This is another reason why the maintenance and implementation of new features is hard to do.
To make it short, i was not willing to reinvent the wheel to get a simple build environment. TinyBSD and all the other build envs for FreeBSD does not meet the requirements needed.
Debian LiveHelper was to solution i was searching for. Also the complete Debian infrastructure seems to reduce the work to get a new FreeNAS to a minimum. The FreeBSD ports system is great, but i do not want to waste my time by fixing build problems, i only want to install and use software. Implemenmting the services and WebGUI is time consuming enough.
My decision to use Linux for the next version was because there are too much bugs in the core FreeBSD system. Simply have a look into the bug tracker. FreeNAS does not run on many systems, mainly new hardware makes trouble. The main reason is the driver problem with FreeBSD which seems to be no problem with Linux because there are great companies in the back that support it. Also the Linux developer community is much greater than the FBSD one.
Finally i have to say that the decision was not easy, but if you have to reimplement something from the beginning (and THIS IS NECESSARY to get rid of all the limitations), then you also use the latest technology that is available at the moment. And at the moment this is Debian/Linux (better driver support/maintenance/package system/docs/FUTURE-PROOF). The only limitation is the missing of ZFS, but i think there will be some equal filesystem available in Linux soon.

The main aim of the next generation is to make it as much as possible enhanceable. This does not mean that FreeNAS will include anything out-of-the-box (i plan to reimplement all the services that are available right at the moment), but it can be enhanced easily by the user via interfaces and scripts that can be copied to the system.

I know that there will be some users that will not be happy about that, but does it make sense to have a cool system that does not run on new hardware?

Regards
Volker

A short list of pros:
- Text and grahical installer that can be customized. This means no hand written install scripts anymore which causes some problems in FreeNAS
- WOL works in Linux
- lmsensor - A WORKING sensor framework which is a really needed feature in FreeNAS to check the CPU/MB temps and fan speeds
- Better Samba performance
- Ability to implement HA features
- System can be updated via 'apt-get' or any other deb package manager
- Better driver support
- Maybe 'ZFS' over FUSE (there is already one commercial product available that uses this feature)
- NFS4
- ...

More in thread here:
http://sourceforge.net/apps/phpbb/fr...t=3966&start=0

Will they also change name from FreeNAS to DebNAS?

[DutchDaemon]

I think it's going to be CoreNAS.
http://sourceforge.net/projects/corenas/

[DNAeon]

Originally Posted by DutchDaemon

I think it's going to be CoreNAS.
http://sourceforge.net/projects/corenas/

Yeah, you are right.

Hm, not very happy about the news, though..

What I'm afraid of are probably questions that most of you have already asked to yourself.

What will happen with:

• Solaris
• MySQL
• ZFS
• Java (not that I like it, though, but anyway :P )
• etc..

Some are talking that Btrfs is going to be the "next" ZFS, and my worries are what will Oracle eventually prefer - will they stop developing ZFS and choose to further develop Btrfs, leaving ZFS behind..?

What about Solaris, MySQL, ...?

Also, from one of the comments that Volker made, I cannot see how a Debian system can provide more security than a BSD one ?!

Anyway, seems like that when the end world comes there will be only a few true UNIX descendants and the other part of the world will be in love with their new dictator MSLinux

Regards,
DNAeon

[phoenix]

Note: I have not used FreeNAS, just going by what I've read about it, mostly in that thread.

Considering a normal install of FreeBSD fits into 512 MB of disk space, and that it's virtually impossible to buy a CompactFlash or USB stick that's less than 2 GB, why switch from using custom, hacked-together scripts for installing FreeBSD to a completely different OS and re-do everything from scratch? Why not just switch to a standard FreeBSD install? Use the pkg_* tools? And just wrap a web GUI around it?

[TerryP]

I agree with phoenix, although I must admit, I don't know all of the architectural limitations that FreeNAS had to be shoe horned into.


The only thing that upsets me, is ' Linux developer community is much greater than the FBSD one', which would have been nicer if phrased clearly in the numerical sense :-P

[gkontos]

My decision to use Linux for the next version was because there are too much bugs in the core FreeBSD system. Simply have a look into the bug tracker. FreeNAS does not run on many systems, mainly new hardware makes trouble. The main reason is the driver problem with FreeBSD which seems to be no problem with Linux because there are great companies in the back that support it. Also the Linux developer community is much greater than the FBSD one.

I have to disagree with all of that. I have used linux a lot in the past (Debian, Centos mainly). When I discovered FreeBSD 5 years ago I was so impressed with the simplicity the stability the flexibility and the security.
I also don't understand the issue with the device drives when we are talking about a simple NAS system. I will not go into the discussion of which community is better. This is completely childish.
I have also see in action software raid in linux and I can't compare it with FreeBSD gmirror!!! Let alone ZFS which is really the future in file systems.

George

[Oliver_H]

ZFS has got a future? Where? Oracle is the initiator of brtfs, so where do you see a future for ZFS? Where do you see a future for Solaris? Oracle is a major contributor to Linux, why should they support any further development in this UNIX-incarnation?

Most of you should be familiar with Mr. Solaris, Ben Rockwood:

http://www.cuddletech.com/blog/pivot/entry.php?id=1089

That said, I'm an avid user of ZFS and I'm looking really forward to its development. But I have a very bad feeling about it.

[DNAeon]

Originally Posted by Oliver_H

ZFS has got a future? Where? Oracle is the initiator of brtfs, so where do you see a future for ZFS? Where do you see a future for Solaris? Oracle is a major contributor to Linux, why should they support any further development in this UNIX-incarnation?

Most of you should be familiar with Mr. Solaris, Ben Rockwood:

http://www.cuddletech.com/blog/pivot/entry.php?id=1089

That said, I'm an avid user of ZFS and I'm looking really forward to its development. But I have a very bad feeling about it.

Yeap, that was exactly what I meant...

[Oko]

Originally Posted by Oliver_H

That said, I'm an avid user of ZFS and I'm looking really forward to its development.

How do you expect ZFS to be developed after the dead of SUN Microsystem?
ZFS is a proprietary product with CDDL license. Who is going to develop it after the mother company is dead? Lets not kid ourself. Sparc is dead and so is Solaris and so iz ZFS although the corp is still warm. The only way that Solaris and ZFS could have survived was if SUN really open sourced them with BSD license as Wasabi did with WAPBL. That move would have cost share holders 8 billion dollars so I think it is silly to compare Wasabi and SUN.

I will repeat my comment from the FreeBSD forums regarding FreeNAS. Any BSD user who needs bunch of PHP programmers to configure various services for him should be asking himself/herself why do you use BSDs to begin with.

As of number of drivers, developer community, and the rest of the comments regarding BSDs and FreeBSD in particular by FreeNAS developers they are mostly right on money. But it is silly from them to compare a huge commercial proprietary operating system like Linux with something which is more of hobby (research at best) Operating System. If that was the goal they should have picked RedHat Linux and build the storage based on it from the very beginning. It is also silly that they are replacing one hobby project (FreeBSD) with another Debian. That just shows total immaturity.

[phoenix]

If the deal to buy Sun goes through, I highly doubt MySQL, Solaris, or ZFS will die. Why? Because they allow Oracle to become the next IBM. They will have full control over the entire computing experience, from the hardware, to the OS, to the storage stack, to the software.

You want to run a storage server? Solaris + ZFS.

Want to run an enterprise database server? Solaris + ZFS + Oracle DB.

Want to run a non-enterprise database server? Solaris + ZFS + MySQL.

Want to run an enterprise web server? Solaris + ZFS + Java middleware + Oracle DB.

And so on.

The only bit that may be dropped in the merger would be SPARC. Everything is heading toward 64-bit x86 hardware. Which is kind of sad considering the technology in the T1/T2 line, especially for web servers.

MySQL will become the entry-level Oracle DB, with a migration path available into the full Oracle.

Btrfs is still a good 3-5 years until it's ready for enterprise storage uses. And the UI for it is just horrible (they're still trying to figure out how to make df work on a btrfs). Not to mention the horrible management tools for it (the "filesystem" is the lowest layer in the stack? Really? With RAID done as part of the filesystem? And the filesystem split into sub-volumes? All managed via mount?). Development won't stop, but it's nowhere near a ZFS-level solution.

ZFS has been under development and in use for 10 years now, and is still under heavy development. It's still receiving features that Btrfs doesn't even have on the radar (dedupe just went in last week, over-the-wire dedupe went in, triple-parity (RAID7?) went in before that with a goal of having N-parity support in the future, encryption is almost ready to go int, etc). ZFS development won't be stopping anytime soon. And even if Oracle tries to drop it, it's open-source, and I highly doubt that the heavy-coders for it are going to just stop working on it.

IOW, Oko, you really need to turn down your hatred toward Oracle and Sun. If anything, you should be railing against the EU for slowing things down to the point that the merger has become a battle of attrition, by which I mean that there won't be anything left of Sun's customer base by the time the EU realises that there's nothing to worry about.

[Oko]

Originally Posted by phoenix

IOW, Oko, you really need to turn down your hatred toward Oracle and Sun.

I do not know where you got that. It breaks my heart to see one after another real technology company going under the water. There is nothing that would make me more happy than to see SUN striving and SGI and DEC coming back out of the grave yard. Trust me. I am just not as optimistic as you are.

Speaking of EU I could not care less what those bullshiters have to say about merger. Between IBM, HP, DeLL, and late SUN; U.S. corporations control 85% of hardware server market. Do not forget that the strongest Japanese manufacturer Fujits-u have no R&D of their own but sells SUN's technology.

So much ado about nothing that comes from EU.

[Carpetsmoker]

The main reason is the driver problem with FreeBSD which seems to be no problem with Linux because there are great companies in the back that support it. Also the Linux developer community is much greater than the FBSD one.

I don't understand why he switched to Linux ... Windows has much more drivers and a much larger developer community!

Anyway, FreeNAS is based on m0n0wall, As soon as I saw that m0n0wall replaced /etc/rc* with a bunch of PHP (PHP! Of all languages!) scripts I lost all interest in it ...

[ephemera]

Originally Posted by phoenix

If the deal to buy Sun goes through, I highly doubt MySQL, Solaris, or ZFS will die. Why? ... the merger has become a battle of attrition, by which I mean that there won't be anything left of Sun's customer base by the time the EU realises that there's nothing to worry about.

Great post. Thanks.

(BTW, guys, when are we getting the 'thanks' button back?)

[Carpetsmoker]

Originally Posted by ephemera

(BTW, guys, when are we getting the 'thanks' button back?)

It's been back ... for over a month actually

[ephemera]

Ha ha, damn, I don't know how I missed the "Thanks" button.

[Oliver_H]

Originally Posted by Oko

How do you expect ZFS to be developed after the dead of SUN Microsystem?
ZFS is a proprietary product with CDDL license. Who is going to develop it after the mother company is dead? Lets not kid ourself. Sparc is dead and so is Solaris and so iz ZFS although the corp is still warm. The only way that Solaris and ZFS could have survived was if SUN really open sourced them with BSD license as Wasabi did with WAPBL. That move would have cost share holders 8 billion dollars so I think it is silly to compare Wasabi and SUN.

I will repeat my comment from the FreeBSD forums regarding FreeNAS. Any BSD user who needs bunch of PHP programmers to configure various services for him should be asking himself/herself why do you use BSDs to begin with.

As of number of drivers, developer community, and the rest of the comments regarding BSDs and FreeBSD in particular by FreeNAS developers they are mostly right on money. But it is silly from them to compare a huge commercial proprietary operating system like Linux with something which is more of hobby (research at best) Operating System. If that was the goal they should have picked RedHat Linux and build the storage based on it from the very beginning. It is also silly that they are replacing one hobby project (FreeBSD) with another Debian. That just shows total immaturity.

>How do you expect ZFS to be developed after the dead of SUN Microsystem?

Well, Oko, I hope that Sun survives. But as I said, I have a very bad feeling about it ;-)

[kwag]

Originally Posted by Oko

It is also silly that they are replacing one hobby project (FreeBSD) with another Debian. That just shows total immaturity.

OMG!
What weed are you smoking, Oko?
FreeBSD a "hobby project" I LMAO!!!

Then to you, the Internet is a "hobby project", because most of it runs on BSD code and NOT Linux!

[roundkat]

I run FreeNAS for a friend of mine's business and
wish the project would stay with FreeBSD, but the point
"I think" is that Volker "wants" to give more features to users by switching..

My friend is not the technical and wants some that just works..
FreeNAS meets his needs..
That being said, he will stay with FreeNAS .7 using ZFS.

His Project -- His Choice -- His Time --
and OpenSource projects are all technically "hobby projects"
If you equate non-paid developers = Hobby Project.
Most OSS devs are not paid and volunteers.. much like this Wonderful forum...

OpenBSD being (IMHO of course) the biggest
"hobby project" since it is "by the developers", "for the developers"

Regarding Oracle /Sun merger..
It all about the bottom dollar.
I think we all understand that Companies buy other Companies to :
- squash competition
- buy technology they don't have or they competitor is better
- implement the newly acquired technology to add more functionality to their existing product..

I believe:
- Sparc sales must have been going down for some time so Sun started developing and using
AMD Opteron x64 architecture to extend the market to the Non-Enterprise level businesses and I personally feel that will be the trend.
- Why continue to make hardware that is essentially Enterprise Level when more and more companies
are moving toward x86 / x64 bit architecture..?

We are a Sun Shop and have boxes that cost 10's of thousands of $$$..
And we pay $$$ for support contracts every year..

I love Solaris but Sparc to me is way overpriced, we could easily move to
a x64 bit platform without any issues and would get more bang for the buck..

The trend of other organizations where I work is a move to Linux..
Why..
-- it meets the needs of the business
-- Support is very reasonable
-- very inexpensive x86 / x64 bit hardware

We should applaud the contribution for Volker continuing this OSS project not criticize him for
not using FreeBSD...

my Thoughts ...

[J65nko]

Very strange is that the FreeBSD website at http://www.freebsd.org/news/press.html#story200911:01 mentions ZFS support in FreeNAS : http://www.h-online.com/open/news/it...rt-853475.html

[vermaden]

@J65nko

The decision has been made after the release of FreeNAS 0.7 with ZFS from what I know.