Weary traveler of 2020, rest here with some soothing, happy tech news. FreeBSD finally merges in OpenZFS

On Tuesday years of work culminated in the integration of OpenZFS, an open-source storage platform, into the code base of the FreeBSD operating system. The union took the form of a code merge committed by Matthew Macy, a senior software engineer at IX Systems who contributes to both projects. "The primary benefit is …

[Anonymous Coward]

It's not the Ubuntu lawyers who need to give their ok

It's Oracle lawyers that must approve that. Yep, same lawyers who are about to extract billions of dollars from Google for few lines of code and who managed to copyright Java APIs. Oh and by the way, Oracle happens to have a Linux business so I would rather not provoke them.

[Anonymous Coward]

'Linux business' ?

Making money off of the backs of the altruistic freely-supplied efforts of others?

[Arthur the cat]

Re: It's not the Ubuntu lawyers who need to give their ok

It's Oracle lawyers that must approve that. Yep, same lawyers who are about to extract billions of dollars from Google

Google has billions of dollars to extract. Open source projects like FreeBSD don't, so any attack on them would only bring (more) ordure down on Oracle's head.

[Len]

Re: It's not the Ubuntu lawyers who need to give their ok

I am not that convinced that Oracle is still looking for some company to sue for the use of ZFS. They are notoriously litigious and they had a chance to sue companies ranging from Apple and Canonical to some major storage vendors and they never did.

[Paul Crawford]

Are Oracle developing ZFS in any meaningful way still?

After the take-over I got the impression a lot of Sun's top technical talent left.

[Anonymous Coward]

Are Oracle developing ZFS in any meaningful way still?

There's still a ZFS team, focussed more on improvements for the storage appliance than just on the filesystem for Solaris.

After the take-over I got the impression a lot of Sun's top technical talent left.

"left" is one way to put it. "Were shown the door" is perhaps more accurate.

[Anonymous Coward]

@Paul Crawford - That is irrelevant

Lawsuits will bring them way more money than engineering.

[bombastic bob]

FreeBSD and "Linuxy" code

often time FreeBSD (and the other BSDs) might lag behind others because developers often ignore their existence and write "Linuxy" code, code that makes assumptions that you're ONLY going to run it on Linux. That's not very POSIX of them, but it happens a LOT.

And FreeBSD has had a good strong working ZFS implementation running for a LONG time now. So porting FROM FreeBSD's implementation would make good sense.

However, two split ZFS code branches appeared instead. One, FreeBSD's ZFS implementation (which I've been using for several years now), and OpenZFS, apparently targeting Linux. But which one got "most of the love"? The 'Linuxy" one, apparently...

Yet, maybe Linux could benefit from some FreeBSD ZFS features, too. (As far as OpenZFS features go, I'm actually pretty happy with how it is on FreeBSD at the moment so "meh").

Here's one thing FreeBSD has done VERY right with ZFS: A few years ago FreeBSD more or less perfected "boot into ZFS" so you could have a pure ZFS system. But if you want, you can still boot UFS and have ZFS volumes as well. Setting the system up for "all ZFS" or UFS boot is pretty easy. In fact, pure ZFS is a bit easier (for new installs).

I've seen references to "boot into ZFS" being done in Linux, but everything I've read so far seems "hacky" and may be "not officially supported" by the distro (or anything else for that matter).

[Len]

Re: FreeBSD and "Linuxy" code

Contribution from FreeBSD’s ZFS implementation is exactly what happened, although not as much as from the Linux implementation.

Essentially OpenZFS 2.0 is a combination of code from Illumos, FreeBSD, ZFSonLinux, iXsystems, Nexenta Systems and possibly more.

FreeBSD contributed its Trim support until Nexenta donated its superior Trim support (it supports queuing as in “Trim this device but not until the system is idle”). FreeBSD also donated its NFSv4 ACL code and its vdev ashift optimisation.

ZFSonLinux contributed the most as their code was most advanced. iXsystems brought their experience in running ZFS on serious storage appliances to the table and is probably the party that will test this the most as they will ship it on production storage systems long before FreeBSD ships it in Stable.

It really is a collaborative effort and quite impressive as far as I’m concerned. Have a look at this talk to see how far it has come from a whole number of different code bases: The Future of OpenZFS and FreeBSD by Allan Jude