Hi all,
We're currently thinking about moving one cluster to proxmox. Storage used will be about 130 TB. I read about and tested PBS, works like a charm. But not on spinning rust.
First question: what compression / dedup rate can I realisticly expect? The 130 TB storage in use are 90% Ubuntu 22.04 VM's.
I would like to re-use some "older" hardware we have laying around. One of the pieces of this would be a whole lot of Samsung PM863a drives. These are fast, but "only" SATA SSD instead of NVME. Would this be good enough? I did some testing, seems OK, but I can't mimic backing up the 130 TB.
When it comes to CPU & memory, what is the recommendation for this?
I would love to use ZFS because of it's feature set and simplicity in expanding/replacing stuff. I'd disable compression and checksumming, basic leaving me with something that does perform closer to ext4. Did someone else test this already?
We're currently thinking about moving one cluster to proxmox. Storage used will be about 130 TB. I read about and tested PBS, works like a charm. But not on spinning rust.
First question: what compression / dedup rate can I realisticly expect? The 130 TB storage in use are 90% Ubuntu 22.04 VM's.
I would like to re-use some "older" hardware we have laying around. One of the pieces of this would be a whole lot of Samsung PM863a drives. These are fast, but "only" SATA SSD instead of NVME. Would this be good enough? I did some testing, seems OK, but I can't mimic backing up the 130 TB.
When it comes to CPU & memory, what is the recommendation for this?
I would love to use ZFS because of it's feature set and simplicity in expanding/replacing stuff. I'd disable compression and checksumming, basic leaving me with something that does perform closer to ext4. Did someone else test this already?