While Oracle hasn't annouced definative plans that seems extremely unlikely unless there is a major shift in thinking. The major difference between Btrfs and ZFS is that the former is targeted at Linux and the latter at Solaris.
I will admit right now that I really don't know much about the creation of new file systems. However, shouldn't that be completely irrelevant? From my understanding, ZFS could work on Linux, however due to licensing restrictions, people are being forced to implement it, at best, in FUSE. And given that ZFS could also work on the Mac platform, I draw the conclusion that a file system originally designed for Linux or Solaris or Mac is irrelevant, it can be ported and made to work with the proper modifications for other UNIX-like operating systems.
Since Larry Ellison said he thought Solaris was a very good implementation of Unix, it seems unlikely that Oracle is going to toss Solaris into the trash can.
I didn't think Ellison would simple throw Solaris into the trash. No point in wasting the work and code behind it.
If Solaris is around then ZFS is most certainly going to be around. For example one of the fastest growing products Sun had was its Unified Data Servers. ZFS is a key component of the functionality of those systems.
The existence of Solaris and ZFS is probable not tied. If we take what I said about file systems and their portability true, then Solaris could survive with Btrf which could become an evolutionary descendant from ZFS, superior to it. Not that I'm saying Btrfs *will* evolve into a superior file system and successor to ZFS, it is just that I think that is the likely case.
I don't think Oracle is going to abandon Linux, but if there is a viable amount of money in Solaris support revenues (seems very likely) not going to kill off Solaris. For sure there will be some battling around killing the other off but think Oracle is large enough can do both. Especially since a huge chunk of Oracle Linux is just Red Hat Linux. ;-)
I'm having trouble grasping how Btrfs won't become a direct competitor to ZFS in terms of overlapping features and interests. Yes, there is enough room for both to exist, however, what would be the point of that? Now that Oracle has the code base of both Btrfs and ZFS, it would make sense to consolidate all the gains in ZFS and combine them with all the new advancements in Btrfs, especially since Btrfs is still in the development stage. ZFS was suppose to be the last word in file systems remember?
Frankly Btrfs looked alot when they announced that they wanted a ZFS clone that was compatible with the Linux/GPL constraints. Throw in Rieser being sent off to jail and even more folks jumped on bandwagon. There are a few things different ZFS but there is a fair amount of overlap.
Which is precisely why there is no reason for ZFS and Btrfs, now being owned by the same company, needing to keep both. It makes sense to simple dual license the currently early in development Btrfs so that it supports both the Linux and other OS markets. Pour in the experience features and some of the code that went into ZFS into Btrfs. Its a win-win for everyone.
Btrfs has compatibility with ext3/ext4 in mind. It is a very, very Linux oriented file system. Moving it to another OS is very unlikely.
I don't see how being compatible with EXT3/EXT4 or being very Linux oriented has anything to do with its portability. Linux is lesser UNIX and should be portable given that Linux itself extends out in many areas. EXT2/3 and I think 4 can be mounted, read-write by the Mac OS. There is just no reason to have EXT3/4 used in place of HFS at this point vs ZFS/Btrfs which are clearly superior and provide definable advantages to Mac user-base.
Also, assuming that Ellison doesn't get rid of Solaris, why develop two overlapping filesystems [which as you say, Btrfs being somewhat of a clone of ZFS] when one will do and can be dual licensed eliminating all other problems with adoption in both arenas? Consolidate the code and resources and developer teams into one FS team.
Similarly, if look at the Automatic Storage Mangament inside the database as another "file system", there are already multiple files systems under development at Oracle. So it doesn't have to be "just one filesystem for the whole corporation" and starve resources off of one so that the other can get ahead.
Your looking at very specific storage needs being handled there. With Btrfs, as you said, being a Linux clone of ZFS, it doesn't make any sense to do that when you actually own the product. You don't need to clone it anymore, you own it and can simple relicense it. Btrfs combined with ZFS would make an even better *general purpose* server/client filesystem solution, which is what ZFS is and what Btrfs is bound to become. The storage solution you listed has very specific applications not suited to what ZFS and Btrfs seem to be aimed at.
Btrfs is useless for Apple to include into the kernel because it is GPL based. No GPL stuff is going inside of the core OS at Apple. No way, no how. Similar reason why ZFS can't going into Linux is the reason it can go into MacOS X.
Which is why you have software that is dual licensed. I don't disagree with your comment about the licensing, its just that there are licensing solutions to licensing problems.
