Currently we're running a PBS server containing 10 Samsung PM983 7.68 TB NVME drives. But.. we're coming to the point where we need to upgrade. Performance is excellent, but limited storage space. I have no extra bays available for more drives. The only option would be to upgrade each drive to a 15.36 TB version of the NVME, but at a cost. Other option would be to do some funky stuff with nvme over TCP.
Another idea came to mind: currenly we hold something like 20 days of retention on the PBS server. What if I drop this to 7 and everything older than 7 days goes to a slower PBS instance. Not sure how I would accomplish this, is there a way to do this in PBS? Or do I need to setup seperate backup jobs on the cluster where both PBS's store the data for the first 7 days and only the slow PBS stores it longer?
Regarding hardware for the slow PBS: would a bunch of 7200 RPM SAS drives, 2 ZIL disks and 2 metadata devices be sufficient?
I'm a little afraid that when I start a backup job we'll saturate the disks since all hosts start at the same time. Would be a nice feature in PVE to be able to limit backup jobs to max X concurrent.
What do you think about a 2 tier PBS solution?
Another idea came to mind: currenly we hold something like 20 days of retention on the PBS server. What if I drop this to 7 and everything older than 7 days goes to a slower PBS instance. Not sure how I would accomplish this, is there a way to do this in PBS? Or do I need to setup seperate backup jobs on the cluster where both PBS's store the data for the first 7 days and only the slow PBS stores it longer?
Regarding hardware for the slow PBS: would a bunch of 7200 RPM SAS drives, 2 ZIL disks and 2 metadata devices be sufficient?
I'm a little afraid that when I start a backup job we'll saturate the disks since all hosts start at the same time. Would be a nice feature in PVE to be able to limit backup jobs to max X concurrent.
What do you think about a 2 tier PBS solution?
Last edited: