ZFS isn’t that resource intensive, although it does like RAM.
But why not just add additional bricks? Gluster is kind of built to use disks as bricks, and rely on gluster replication to provide redundancy in the data. Either replication or distribute-replication can provide protection against disk failures.
Em ter., 28 de jul. de 2020 às 17:08, Alvin Starr <alvin@xxxxxxxxxx> escreveu:
Having just been burnt by BTRFS I would stick with XFS and
LVM/others.
LVM will do disk replication or raid1. I do not believe that
raid3,4,5,6.. is supported.
mdadm does support all the various raid modes and I have used it
quite reliably for years.
You may want to look at the raid456 write-journal but that will
require an SSD or NVME deivce to be used effectively.
On 7/28/20 3:43 PM, Gilberto Nunes
wrote:
Hi there....
'till now, I am using glusterfs over XFS and so far so
good.
Using LVM too....
Unfortunately, there is no way with XFS to merge two or
more HDD, in order to use more than one HDD, like RAID1 or
RAID5.
My primary goal is to use two server with GlusterFS on top
of multiples HDDs for qemu images.