4 ms·
The post is literally about how ZFS compression saves them millions of dollars.
by whitepoplar 5y ago
The post is literally about how ZFS compression saves them millions of dollars.
- pengaru 5y ago> The post was literally about how ZFS compression saves them millions of dollars. ... relative to their previous ZFS configuration. They didn't evaluate alternatives to ZFS, did they? They're still incurring copy-on-write FS overhead, and the compression is just helping reduce the pain there, no?
- drob 5y agoZstandard gets us 5.5x compression. The previous ZFS config got us 4.4x compression. XFS, which we ran on for years before rolling out ZFS, does not compress.
- pengaru 5y agoThanks for the clarification
- linuxandrew 5y agoThere aren't many good alternatives to ZFS if you specifically want snapshots and compression. LVM? Managing it sucks and the performance is bad (there's plenty of information on this). I think Red Hat's Stratis was supposed to improve on some of this but it's not mainlined yet. For a while it sounded like Red Hat were going down the device-mapper functionality route of Stratis, VDO, dm-integrity and XFS, but I don't know where this is at. Btrfs is probably the closest bet, but it's software RAID 5/6 has the known write hole bug. Still waiting for bcachefs to get mainlined. ext4 doesn't do CoW. XFS kinda does with its reflink functionality but it doesn't do filesystem-level snapshots.
- rolandog 5y ago> Still waiting for bcachefs to get mainlined. Me too! As well as waiting for the successful launch of the James Webb Telescope...
- namibj 5y agoThe Raid5/6 might not matter, if they are running on AWS with local NVMe storage, as defect hardware will kill the VM anyways, and how often do they get a read checksum error?
- effie 5y ago> LVM? Managing it sucks and the performance is bad (there's plenty of information on this). Can you give some recent references to that claim? With thin LVM (introduced in rhel6-rhel7), performance is supposed to be OK.