3 ms·
I use FreeBSD on servers instead of Linux for one reason: ZFS. It's really that much better than anything that is currently available on Linux. BTRFS is not e
by rubyist5eva 4y ago
I use FreeBSD on servers instead of Linux for one reason: ZFS. It's really that much better than anything that is currently available on Linux. BTRFS is not even close.
Yes I know it's available on Ubuntu but everything else about Ubuntu is just so messy.
- stjohnswarts 4y agobut you can use ZFS on linux?
- rubyist5eva 4y agoYes but it’s clunky. ZFS on FreeBSD is smooth as silk. Ubuntu is the only distro that I would say has good integration with ZFS but I don’t like anything else about Ubuntu.
- cyberpunk 4y agowhat linux server distro are you using if not ubuntu lts? Im not a massive fan either but i run thousands of the bastards (k8s hosts) and cant imagine an other... just disable snap and you're good ;)
- rubyist5eva 4y agoAt $CURRENT_JOB we run our postgres database on a massive bare-metal dedicated server running FreeBSD, our container hosts are Alma Linux 8 and the base images are typically debian-slim, though some are alpine or even scratch.
- Asdrubalini 4y agoI think NixOS has an even better ZFS implementation, compared to Ubuntu.
- School-Cotton 4y agoI don’t have a lot of SRE/sysadmin/etc. experience, so this is probably just ignorance on my part, but I find it hard to imagine a scenario where the filesystem on my servers matters at all (as long as we’re not talking about something super ancient like FAT). What are some workflows that ZFS enables for you on servers that you couldn’t do with ext4 or stock FreeBSD FFS ?
- CoolCold 4y agoPretty much * data compression - say midsized 1.6TB mysql db compressed into 400Gb => allows to run replica servers on cheaper hardware (smaller drives) * hot data set caching - depending on case can be 5-100 times better performance * single pool but different settings per FS - say block sizes 16k for DB and 256kb for file storage * snapshots. Instant snapshots * coming back to db replication case, this helps to test db scheme migration on close to production env and prepare better planning/downtime window. Once tests done - rollback snapshot and enable replica back without full reclone. And so on