OpenZFS 2.0 Out In 2020 With Unified Linux/FreeBSD Support, OpenZFS 3.0 With macOS

[aht0]

Poor you BTRFS...
Hey Bcachefs, I wonder.. how you doing?

10 years later:
Poor you Bcachefs..
Hey NewShinyFS, how you doing?

[kreijack]

Sorry just my usual unqualified lament
I just see btrfs development stagnant and zfs getting slowly traction in the Linux ecosystem by offering a complete solution. I think btrfs might have many technical pros over zfs (on the paper), but I think Oracle behind the scenes it's fuckin it all up...

[Disclaimer, I made some small contribution to BTFRS]
I have to agree that it seems BTRFS lost some attraction. However I have to point out that recently in the BTRFS mailing list appeared some patches like:
- RAID1 with 3 and 4 copies
- A more smarter chunk allocator (with handle better the degraded mode)

So I don't think that talking of "stagnant development" is correct. Anyway I am guessing which filesystem could be considered under fully development: even XFS, EXT4, BTRFS, OpenZFS/ZOL are in a consolidating phase.
The only exception which comes to me is bacachefs (!) ; however this is a one-man project; I was never able to find any mailing list; no community ... So I don't have too much expectation.

My feeling was that BTRFS is a very complex beast.
Collapsing the layers filesystem and device management in the same thing, at the time seemed a great idea (a lot of opportunity for improvement, like RAID rebuilding of the basis of the checksums) however create a very large code base with a lot of nasty corner case (like shrinking a filesystem with quota)...

It is true, some features still needed a lot of care (like RAID5/6). I hope that these issue will be addressed in the future.

[Volta]

Your beliefs are not relevant, Sun was quite a bit more than just ZFS. They had a full software stack with Java, OpenOffice, Solaris, down to their own CPU design for their workstation and servers.

They had a huge blow on hardware sales when the dot-com bubble burst, and they never really recovered from that.

Except Solaris was being replaced by Linux before Sun's problems.

[starshipeleven]

That's true. However, if it was good Sun would have done something about it to not let it die. That's my simple point. Maybe it's too simple, but I believe it's true.

Your beliefs are not relevant, Sun was quite a bit more than just ZFS. They had a full software stack with Java, OpenOffice, Solaris, down to their own CPU design for their workstation and servers.

They had a huge blow on hardware sales when the dot-com bubble burst, and they never really recovered from that.

That's yet another reason Microsoft weathered the debacles of Windows 8 and later relatively well, massive loss of hardware sales does hurt them, yes, but not anywhere near as bad as it hurts Intel. So they have time to move to "cloud stuff", while Intel was stuck at trying to teach laptop manufacturers how to copy Apple properly with the whole Ultrabook thing.

Furthermore, there seems to be no interest in open source Solaris.

That's what usually happens to a opensource project that is mostly a code drop (or singlehandedly developed) from a single company when the company dies or abandons it.

You can observe the same behavior with Ubuntu Touch (Canonical's dead smartphone OS), a "community" project is keeping it on life support, but does it really have a bright future?

This is what would also happen if Google decided to just drop Android tomorrow. Most OEMs really are just hacking around Google's code drops, they will just keep it on life support like the Illumos has done for Solaris.

[Volta]

That's just proof of Oracle not wanting to invest time in developing their own OS when they can just milk the old customer base that is locked in at no real cost, especially now that they can just rebrand CentOS as their "way forward".

That's true. However, if it was good Sun would have done something about it to not let it die. That's my simple point. Maybe it's too simple, but I believe it's true. Furthermore, there seems to be no interest in open source Solaris.

[starshipeleven]

I read this article as well, but it was draft that was never finished, because then it would be verified. That article was full of bullshit and Solaris code was just a huge mess. Proof? It's nearly dead and stagnates since long time like 'true' Unix.

That's just proof of Oracle not wanting to invest time in developing their own OS when they can just milk the old customer base that is locked in at no real cost, especially now that they can just rebrand CentOS as their "way forward".

[starshipeleven]

To quote Bryan Cantrill, "What you think of Oracle is even truer than you think it is."

"Do not fall into the trap of anthropomorphizing Larry Ellison"

Lol. that's around 38:00 timestamp in that video. It's good advice, and a fun commentary on Oracle and the death of Solaris.

He always gave me strong Tony Stark vibes.

[Volta]

I wonder if Oracle has any competent people working for it. They have a real Unix OS that pioneered a lot of things, Dtrace, ZFS, Containers and are willing to just let it stagnate. I once read an article comparing the source code of Solaris to Linux and it said how much better engineered the Solaris code was. As an undergrad we got to install OpenSolaris and it was a rock solid system. It pains me to see Oracle not give a darn about it. Instead there best effort is a copy of Red Hat Linux? Really?

I read this article as well, but it was draft that was never finished, because then it would be verified. That article was full of bullshit and Solaris code was just a huge mess. Proof? It's nearly dead and stagnates since long time like 'true' Unix. I thought tracing tools, file systems and containers were available before, so they don't seem to be pioneered by slowlaris.

https://www.phoronix.com/scan.php?page=news_item&px=Solaris-Old-C-Code-To-Python

[starshipeleven]

I just see btrfs development stagnant and zfs getting slowly traction in the Linux ecosystem by offering a complete solution.

I don't disagree. It seems Btrfs has reached the point where the corporate overlords are fine with it (for example the whole "write hole" thing in RAID5/6 mode that indeed still exists is irrelevant for most servers because you have an UPS anyway), and it's too complicated for most lowly consumer peasants to develop the features they want themselves.

Meanwhile, ZFS's core functionality was already developed back in the day when corporate overlords would actually want something like that. Now it is already mostly done so it's just a matter of porting stuff over. I mean their "core development" speed isn't fast either. Getting basic stuff like shrink support (even offline) took them like... 4 years? 5 years? I don't even remember anymore.
Sure they add some special features every now and then but it's always something relatively easy. They still don't have a working defrag, and yes fragmentation can become an issue on some workloads

I think btrfs might have many technical pros over zfs (on the paper), but I think Oracle behind the scenes it's fuckin it all up...

Thankfully, Oracle isn't doing much on either fronts (which is a very good thing). They have some token developers on btrfs and they do honest work on bugs and such by looking at the mailing list, but that's more or less it.

As I said, this is exactly how we want Oracle to remain, very laid back on the software/technical side of things so we can live in peace, while they focus on what evil corporate businessmen do best, license trolling for their current products (like that sending bills for Virtualbox extension usage to an ISP company that does not even have Virtualbox at all), milking money from Google with dubious legal claims about API copyright, and screwing their businness customers by selling them a rebranded RedHat product.

[horizonbrave]

what's wrong with btrfs

Sorry just my usual unqualified lament
I just see btrfs development stagnant and zfs getting slowly traction in the Linux ecosystem by offering a complete solution. I think btrfs might have many technical pros over zfs (on the paper), but I think Oracle behind the scenes it's fuckin it all up...

But I'm just a non-technical, I should refrain from commenting sometimes
Thanks!

PS
wrote you a PM