Comment by blibble
6 years ago
I'll give up Linux on my servers before I give up ZFS
especially so given the recent petulant attitude that broke API compatibility in the LTS branch just to spite the ZFS developers: https://news.ycombinator.com/item?id=20186458
compete honestly on technical merit, rather than pulling dirty tricks that you'd expect of Oracle or 1990's MS
Pretty much my view as well. If Linux becomes incompatible with ZFS in any way, I'll switch to FreeBSD.
That said, after the Oracle Java debacle, I can see why Linus would not be receptive towards merging ZFS into the kernel. I just wish he argued the point on legal issues alone instead of making up stories about non-existent technical flaws in ZFS. The whole thing is basically a work of art. Oracle should consider GPL-ing it and integrating it into Linux directly.
I have been using FreeBSD for its better ZFS support for years, and it's great. Highly recommended.
Same here. I immediately ditched Solaris (hated it) for Linux when KQ Infotech ported ZFS over about a decade ago, but after a few years switched all my file servers to FreeBSD. ZFS > Linux. Did I mention I love Linux?
FreeBSD has been using the ZFS on Linux code for a while.
2 replies →
> Oracle should consider GPL-ing it and integrating it into Linux directly
I think Apache (or BSD/MIT) would be far more palatable, as GPL'ing it would cut off the BSDs as well as OpenSolaris, which would certainly be a bummer.
Por que no los dos? It's their IP, they can license it under several licenses to maximize adoption. This being Oracle though, I don't think it'll ever happen.
> If Linux becomes incompatible with ZFS in any way, I'll switch to FreeBSD
Isn't FreeBSD now using ZFSOnLinux project as well?
no. see above
Yeah...I use FreeBSD for file servers because I don't have to even pay attention this constant ZFSonLinux drama. I treat them like almost like appliances. Linux servers are more than happy to use them on the back-end.
how do you structure that? block images served by iscsi? shares served by nfs/samba?
All of the above, depending on what I'm trying to accomplish. Usually Samba for Windows clients and general bulk storage (pretty much everything can do CIFS mounts at some basic level these days), NFS for *nix and VMWare clients (esp where I can leverage NFSv4), iSCSI for various block image needs, scp/sftp/rsync. All of this is basic out of box for FreeBSD.
In operations, I treat them as semi-black-box (grey-box?) appliance where they only do the file storage function and are moral equivalents of Network Appliance NAS boxes. I don't try to convince the Linux or Windows teams to migrate other workloads to FreeBSD, and mostly don't want them to since that would mean Yet Another app environment to support.
FreeBSD has a really efficient network stack, so I can attach the FreeBSD stores at 10GB (usually, 2-4 bonded 10GB links) and it keeps up fine. I I have colleagues who are doing the same with 40GB links (40-160BG aggregate) to backbone networks, and apparently there are many shops hooking up FreeBSD with 100GB links. Limiting factor seems to be ZFS and the storage subsystems supporting it, not network, which is interesting, but I don't have the ability to benchmark the 40GB and higher stuff.
The core development team hasn't really been fully trustworthy since they spent years pretending their cpu scheduler wasn't hot garbage for desktop usage, denied the need for a pluggable scheduler to allow multiple schedulers to be selected from, then seemingly an age later implemented something in the same vein as CK while giving zero credit.