The Journey from BTRFS to ZFS

The only problem with ZFS on Linux that prevented me from using it in the past was that it should be built as an external module because of licensing issues (Linux's GPL vs ZFS's CDDL) . That was not good for me as I upgrade the kernel to the newest -rc every Sunday . Unfortunately, I was blind to the awesome tool called DKMS . One day , I was reading the ZFS article on Archwiki , and I found that there is this package (zfs-dkms-git) which has a PKGBUILD that is not tied to a specific kernel release . Installing this package was painless using (yaourt -S zfs-dkms-git --noconfirm). It was successfully built againt 4.6-rc5 as of last week , I just had to install the (mkinitcpio-dkms) package, edit (/etc/mkinitcpio.conf) and include the dkms and zfs hooks . After that I issued (mkinitcpio -p linux-mainline) to rebuild the initramfs, rebooted the machine and looked through the dmesg log and this was there :

SPL: Loaded module v0.6.5-49_g224817e
SPL: using hostid 0x000af001
ZFS: Loaded module v0.6.5-219_gda5e151, ZFS pool version 5000, ZFS filesystem version 5

I will start migrating my 7TB data from BTRFS RAID10 to ZFS RAIDZ or maybe RAIDZ2 and see how that would perform compared to the last time I tested ZoL with my 7TB of data 6 months ago .
BTRFS has this "micro" latency that just refuses to go away . I'm accessing this BTRFS RAID10 as an SMB share and it has +7TB of data and there is a noticeable latency when opening this share on my Windows 8.1 (using a Virtio 10Gbps pipe) . I didn't experience the same latency with ZFS holding the same data about half a year ago .
On a side note, BTRFS development was -in my opinion- killed by Facebook. No new ground-breaking features where introduced in the past year as far as I know , only bug fixes and little performance improvements .