ZFS and the future.

[roasted]

I was recently exposed to FreeBSD and FreeNAS and I'm liking what I see. I'm a heavy Linux user but I can definitely see FreeBSD balancing things out in certain areas where I could use them. That said, I began to do a little homework. I read about FreeBSD, its history, pros, cons, and I also caught a glimpse of ZFS being talked heavily about. Further research revealed this was a product of Sun, which unfortunately became part of Oracle with the buyout.

That said, I have some questions. I read that ZFS was closed off, as far as the source code goes. Is there still active development with ZFS, perhaps an open source/non-Oracle version or something like that, or is it something that will have a static lifestyle on its current source code prior to its closing and have little/no change? Are there any FreeBSD developers that also actively work on ZFS?

I suppose I'm just curious about the stability of ZFS in regard to its existence, not so much its implementation. Any other information to fill in the blanks would be great.

Thanks!

 

[vermaden]

That said, I have some questions. I read that ZFS was closed off, as far as the source code goes. Is there still active development with ZFS, perhaps an open source/non-Oracle version or something like that, or is it something that will have a static lifestyle on its current source code prior to its closing and have little/no change? Are there any FreeBSD developers that also actively work on ZFS?

Illumos/OpenIndiana developers have currently bigger issues to resolve (like porting various pieces of OS from FreeBSD or other BSDs that were not open sourced at OpenSolaris times), but as things will settle down there, I do not see anything against further development.

FreeBSD team is also currently busy with polishing latest ZFS v28 on 9-CURRENT to make it as reliable as possible for 9.0-RELEASE which would happen somewhere between 2011.06 and 2011.11 hopefully.

Also Oracle promised (no that I trust in anything they say) that they will release Solaris 11 (not Express) code when its release, so ZFS code could be filled to the latest version.

 

[carlton_draught]

That said, I have some questions. I read that ZFS was closed off, as far as the source code goes. Is there still active development with ZFS, perhaps an open source/non-Oracle version or something like that, or is it something that will have a static lifestyle on its current source code prior to its closing and have little/no change? Are there any FreeBSD developers that also actively work on ZFS?

There already is development. Look at what Pawel has done with GELI. In addition to porting ZFS, he has implemented encryption for FreeBSD (GELI). This is without waiting for ZFS encryption (which we may never see as it is still closed). So I would say that it is highly likely that work will continue given that it is already going ahead. Look at what we already have! Self-healing, guaranteed data integrity (providing you provision your systems with redundancy), filesystem transfers that ensure end to end data integrity, deduplication in v28, the simplicity that is using zfs... what more do we really need? ZFS as is, is already a killer app.

I see that as more and more people catch on to the advantages of ZFS, it is going to really snowball for FreeBSD and its ZFS implementation. For anyone concerned about reliability, ZFS is a no-brainer. You have the potential to eliminate and self-heal errors from storage and memory (using ECC memory). The only thing that really remains is some redundancy to catch CPU errors. But it is way ahead of the what we used to have. Prices have fallen enough in hardware that paying a bit extra in what is a really small expense is now worth it as insurance.

ZFS is now a proven filesystem. You are not chancing your data on it, as you would be with vaporware Linux also-rans that will have to have the users be the guinea pigs. And ZFS has had a few years for the infrastructure to catch up with it, to really flesh out the promise it originally once had.

Smart early adopters in diverse areas of the internet are catching on. I regularly see people talk about their experience with ZFS on forums like overclockers, anandtech, silentpcreview, hardforum, etc. Those are not FreeBSD or Oracle strongholds. They are just smart early adopters. Where early adopters go, the market follows. Where the market follows, so does the money. Where money goes, developers start getting paid. When developers get paid, better software gets built.

Jump in with both feet.

 

[roasted]

Well that sounds a little more comforting. I was a big preacher of OpenOffice... because at work where we have thousands of computers, we don't need any MS Office features, so I was preaching to use OpenOffice since we could save about 100 thousand on Office licensing alone... When Oracle took over and tanked it, I felt like a moron having to switch gears to LibreOffice. Although, LibreOffice so far has been working out better for us and once I explained the story of what happened with Oracle's involvement, the rest of the IT staff at least understood.

That said, it does scream the power of open source. If you are running an open source project and you screw it up, the rest of the world can say "I'll take that, thank you" and continue on with it by simply ignoring you're still existent. That's basically what happened when The Document Foundation came and ported to LibreOffice. The OpenOffice name, in my opinion, will die off in the wake of LibreOffice's rising.

That also sparked the idea with ZFS as well. Here I am looking for a solid NAS operating system and I find FreeNAS. ZFS this, ZFS that. Oh shit... ZFS is owned by Oracle. This certainly isn't good. But it's good to know the code is currently still open so users can at least get the job done and be current.

However, it just made me wonder what kind of people the FreeBSD/FreeNAS team are, along with Illumos, Nexenta, etc. Are they people who can support a file system along with an operating system? Is it something that will continue to shine even though ZFS has all of these great features nobody else has? It just made me wonder if ZFS is this amazing thing that will stay dormant and idle in development due to Oracle's closing of the code... as if there could be a day where Oracle's ZFS is the golden ticket and the open source ZFS that we all know and love will just be inferior to it. Or, on the flip side, are the FreeeBSD/NAS/Illumos/Nexenta developers people who would stand by ZFS and manage it on their own and do the right thing by keeping it an open project?

These are all questions that float around. And all of them could have been avoided if Oracle just didn't act like... Oracle.

 

[phoenix]

This has been mentioned on the zfs-discuss mailing list in this thread.

Nexenta, Illumos, OpenIndiana, FreeBSD, and Linux-related devs are working together to advance development of the opensource version of ZFS. No matter what Oracle does, ZFS development will continue.

 

[carlton_draught]

That said, it does scream the power of open source. If you are running an open source project and you screw it up, the rest of the world can say "I'll take that, thank you" and continue on with it by simply ignoring you're still existent. That's basically what happened when The Document Foundation came and ported to LibreOffice. The OpenOffice name, in my opinion, will die off in the wake of LibreOffice's rising.

This to me is a very important reason why I choose FOSS wherever possible. It keeps you from being in a situation where you are screwed.

However, it just made me wonder what kind of people the FreeBSD/FreeNAS team are, along with Illumos, Nexenta, etc. Are they people who can support a file system along with an operating system? Is it something that will continue to shine even though ZFS has all of these great features nobody else has? It just made me wonder if ZFS is this amazing thing that will stay dormant and idle in development due to Oracle's closing of the code... as if there could be a day where Oracle's ZFS is the golden ticket and the open source ZFS that we all know and love will just be inferior to it. Or, on the flip side, are the FreeeBSD/NAS/Illumos/Nexenta developers people who would stand by ZFS and manage it on their own and do the right thing by keeping it an open project?

I know at least Pawel is competent enough to port ZFS and develop something like GELI, so there is at least him involved. That enough gives me confidence. There are also the people who handle UFS and would be assumed to have filesystem competence.

These are all questions that float around. And all of them could have been avoided if Oracle just didn't act like... Oracle.

I've had good dealings at least with one individual at Sun, who is still with them now they are at Oracle. It's true that Sun couldn't keep going the way they were going. They have to start getting funding for what they were doing, somehow. So I can understand that they need to play a little more hardball now and stop being a doormat. Otherwise the engineers won't have a job, eventually.

I think it is important to also not bite the hand that feeds, and maintain as good as possible relations with Oracle. Look at what we have already. A fantastic revolutionary filesystem that was given to us gratis. Let's be thankful for it. Internally, we can prepare for the worst but if we look for ways in which both FreeBSD and Oracle can benefit from collaboration where possible, we will both be better off.

I think that Oracle may be able to use FreeBSD as a "gateway drug" so to speak. Solaris does have extra RAS features that FreeBSD doesn't. As long as it maintains a drip feed of features to FreeBSD that keep us ahead of the rest, it can both gain adherents to the drug of reliability, while keeping its own closed source offering as something that is easy to migrate to, has increased features, etc. In this way, those firms that are successful and desire some "insurance" of extra reliability etc. can migrate to something that does not cost them much compared to the benefits it offers.