They still can't seem to get BTRFS working anywhere nearly as well as ZFS on FreeBSD. Plus, you get a lovely init structure with no Systemd garbage. I love it.
I wish Linux would separate the BTRFS development tree from the production tree. Then work for stabilizing Mirroring and other current normal features of BTRFS. Use the development branch to add new features like the forever pending RAID-5/6.
A long time ago, (2011?), I used BTRFS as my root file system for 2 of my 4 Linux computers. Not for most of the features. My main goal was to get sub-volumes, so I could perform OS updates on a writable snap shot. For me, that worked perfect and the times I had to backout a Gentoo update, it was not a killer. (My prior scheme was dual root partitions that I alternated updating. Before that, occasional full restore!).
But, BTRFS never stablized. People were loosing data, or having serious mount problems, (fortunately not me). I saw BTRFS going into wonderland of compression, encryption and copy-on-write, (reflink), for individual files. That sounded like a nightmare to manage and maintain. The biggest reason I hated BTRFS, was that I could not tell sub-volume sizes using simple tools, like "df -h". Quota limits would have also been nice, but sizes were more important for me at the time.
So, in 2014 I started looking at ZFS on Linux. Implementing it on one of my Linux computers, (a minature media server), worked great. Got my sub-volume sizes and quota issues 100% resolved. Plus, got some great features, like LZ4 compression, (which at the time BTRFS did not support compression). And a reliable development group, releasing updates every few months. Now for years, all my Linux computers use ZFS. Have not lost 1 byte, (and I run monthly scrubs to verify), to the file system. Perhaps a native Linux file system would be faster, but I want my sub-volumes, (aka ZFS datasets).
With OpenZFS used and developed on 4, (soon to be 6 or 7), OSes, I feel more confident that it will be more reliable, (even on Linux), than BTRFS or possibly B-Cache.
(And for those wondering, the OpenZFS Tier 1 OSes are FreeBSD, Illumos and similar OpenSolaris derivitives, and Linux. MacOS might be considered Tier 1, but also could be Tier 2 since little development is performed on MacOS in regards to OpenZFS. The newer ones are OSv, MS-Wiindows and possibly ReactOS, under development to support OpenZFS.)
Linux still playing catch up. (Score:5, Insightful)
Re:Linux still playing catch up. (Score:1)
A long time ago, (2011?), I used BTRFS as my root file system for 2 of my 4 Linux computers. Not for most of the features. My main goal was to get sub-volumes, so I could perform OS updates on a writable snap shot. For me, that worked perfect and the times I had to backout a Gentoo update, it was not a killer. (My prior scheme was dual root partitions that I alternated updating. Before that, occasional full restore!).
But, BTRFS never stablized. People were loosing data, or having serious mount problems, (fortunately not me). I saw BTRFS going into wonderland of compression, encryption and copy-on-write, (reflink), for individual files. That sounded like a nightmare to manage and maintain. The biggest reason I hated BTRFS, was that I could not tell sub-volume sizes using simple tools, like "df -h". Quota limits would have also been nice, but sizes were more important for me at the time.
So, in 2014 I started looking at ZFS on Linux. Implementing it on one of my Linux computers, (a minature media server), worked great. Got my sub-volume sizes and quota issues 100% resolved. Plus, got some great features, like LZ4 compression, (which at the time BTRFS did not support compression). And a reliable development group, releasing updates every few months. Now for years, all my Linux computers use ZFS. Have not lost 1 byte, (and I run monthly scrubs to verify), to the file system. Perhaps a native Linux file system would be faster, but I want my sub-volumes, (aka ZFS datasets).
With OpenZFS used and developed on 4, (soon to be 6 or 7), OSes, I feel more confident that it will be more reliable, (even on Linux), than BTRFS or possibly B-Cache.
(And for those wondering, the OpenZFS Tier 1 OSes are FreeBSD, Illumos and similar OpenSolaris derivitives, and Linux. MacOS might be considered Tier 1, but also could be Tier 2 since little development is performed on MacOS in regards to OpenZFS. The newer ones are OSv, MS-Wiindows and possibly ReactOS, under development to support OpenZFS.)