Future of ZFS ...

[rhdinah]

Now that Oracle has apparently decided to move future ZFS development from being open to closed and proprietary source ... does anyone know where we are going from here? Will there be a FLOSS team taking the last snapshot of the source forward? It's too bad that this decision was made ... I was hungry for that block pointer rewrite functionality that was promised on the todo list.

Any inputs would be appreciated!

 

[Joshua Parker Ruehlig]

Now that Oracle has apparently decided to move future ZFS development from being open to closed and proprietary source ... does anyone know where we are going from here? Will there be a FLOSS team taking the last snapshot of the source forward? It's too bad that this decision was made ... I was hungry for that block pointer rewrite functionality that was promised on the todo list.

Any inputs would be appreciated!

I read somewhere FreeBSD 10 would start to branch the opensource zfs (version 28), might want to google around a bit as I'm lazy right now =]

 

[rhdinah]

I read somewhere FreeBSD 10 would start to branch the opensource zfs (version 28), might want to google around a bit as I'm lazy right now =]

I finally chose to google it ... although I'm not sure why I didn't before. Maybe I did but used the wrong search keywords.

From what I found:

• Many of the hits show the late 2011 time frame ... and much of it IMHO is hearsay ...
• Supposedly the Illumos/OpenIndiana work has forked the ZFS code at the v28 point.
• There is some interesting hearsay ... most interesting quote:
  
  

  We won't know until after Oracle releases Solaris 11 whether or not
  they'll live up to their promise to open the source to ZFSv31. Until
  Solaris 11 is released, there's really not much point in debating it.

  Well Solaris 11 is out ... and where is the documentation anchoring this promise?

I tested both the last OpenSolaris and OpenIndiana in hopes of moving in that direction. Interest ceased after what seemed a ten minute boot.

Many of the hits praise btrfs as the future file system. With quality implementation years off I doubt whether any serious and competent person would seriously commit their data to this alpha code.

Don't mind me ... I'm still bewildered! If anyone has something else to add ... please do.

 

[matsim]

I have something to update - if you are interested in :)

Basically from what I know lots of the main ZFS developers that left Sun-Oracle have joined companies developing their products based on illumos, the open source continuation of the OpenSolaris ("OS/Net") source code.
OpenIndiana, SmartOS and so forth are distribution based on this code
Oracle has ceased to open source any code related to their Solaris product, that is ZFS too, no code was openly released since the snv147 tag back in 2010.
There seems to be leaked S11 code floating as torrent, but that was considered legally poisoned by the illumos community.

Since there is no more centralised, single-company steered development of ZFS, monolithic version don't make sense anymore, that is why people arounds the open ZFS code are heavily working on the concept of feature flags which allows adding new features independently from each other while maintaining compatibility and interfaces clearly defined. There is also some interesting development going on about libzfs2 and the ZFS testing suite.
On illumos there have been quite some ZFS fixes concerning stability and performance so far, the development is definitly ongoing.

Oracle's ZFS development is now entirely behind their walls and basically they have forked it to their ideas and have not cared keeping compatibility with things after v28.
illumos.org has continued to become the place for the open ZFS implementation where also FreeBSD devs like pjd bring in their patches for issues they discovered on FreeBSD. So no, they will be branched of ZFS somehow with feature flags making development more flexible
It could even be that Oracle may be participating in the "ZFS working group" where illumos and FreeBSD developers also talk to each other.

As such it seems that both ZFS and DTrace which are both available to FreeBSD definitley kept and keep kicking without Oracle's contribution - both illumos and FreeBSD devs are working together where they feel it makes sense to them to keep these technologies going.
BTRFS may be a supported FS in Oracle's OEL and SUSE SLES today but it's still deemed heavily unstable for production in mainline Linux as of today.