6 ms·
It's quite the opposite in fact. BTRFS was pulled INTO the kernel recently. It's seeing massive development and is production-ready. > Why not ZFS? The main r
by garritfra 5y ago
It's quite the opposite in fact. BTRFS was pulled INTO the kernel recently. It's seeing massive development and is production-ready.
> Why not ZFS?
The main reason for me is its flexibility. Just an example:
Expanding BTRFS is just a matter of shoving in arbitrary disks and rebalancing the existing array. ZFS requires you to plug in drives of the same size to properly scale.
I wrote about this some time ago, in case you're interested:
https://garrit.xyz/posts/2021-02-07-storage-setup https://garrit.xyz/posts/2021-02-07-storage-setup
- thghtihadanacct 5y agoI suppose Ive been using ZFS in production for so long and its been so stable that anything that isnt fully distributed wouldnt be a step up for us. CentOS8 had ZFS in the kernel, I think debian still does (ubuntu file servers using AD and SMB does not work well in my experience tho), so it looks like im going to alma or rocky. I guess Ill figure out which file system we will use at that point depending on native support. There are other ways to scale ZFS too but I have also found trying to mix and match drives, machines in a production environment is going to bring its own issues.