Top positive review
96 people found this helpful
20W zfs box under my desk? Yes, please!
on March 31, 2014
Once you go ZFS, you can't go back. One word: checksums (and transaction groups, i.e. no fsck/file system check is required ever). If you have family photos, important documents, movies, or music, bitrot and data corruption is a silent enemy. And not something you ever have to worry about with ZFS (with checksums enabled, the default setting). The first time you see the data corruption counter tick up out of 0, you'll wonder how you survived until now - mental it's a game changer to realize this happens periodically (and sometimes on critical files).
Like I said, I've been using ZFS for years in production and I'm ecstatic to have a low power unit for my home office (I purchased my unit directly from iX once this new form-factor came out because it has dual GE NIC ports, which means I can replace my router with this box). 20W, 16GB of RAM, up to 12TB usable in a RAID 3+1 config and able to saturate a 1GE link. Excellent.
Other low end Linux or embedded NAS units are infuriating when you need to troubleshoot something or there is some kind of corruption. This is a small, low power PC that you can salvage replacement components out of and even transfer your drives to a stock FreeBSD host. The OS is mounted via a RAM disk which is loaded from a read-only internal flash drive. Also, with 12TB of usable storage (pre-compression), not having to file system check ever is a big deal if the power goes out. File system transaction groups are sync'ed to disk every 5 seconds my default. If you're paranoid, count to 30.
Last bit of new owner advice: don't forget to setup a monthly ZFS scrub to detect early drive failures or corrupt bits. With a RAIDZ in a 3+1 configuration, ZFS will automatically correct corrupt bits. On this unit I run with sync=disabled because there aren't any documents where I work where I couldn't possibly loose the last few seconds of data in the event of a power outage, but the performance tradeoffs for day-to-day use will add up (if this was a production host with critical data where I couldn't loose even 5 sec worth of writes, then I would run with sync=standard). For a specific directory that does have critical files, create a new zfs dataset and change that one specific dataset to use sync=standard (or sync=always if you're hyper paranoid about possible data loss in the next 5 seconds).
Err.. and this. I'm a power user and am excited by this unit because the OS software is best in class, even in the enterprise space. BUT!, this unit is easy to use from the GUI and novices can easily use 95% of the box's primary functionality. The other 5% is what I'm excited about because I can SSH in to the host and do whatever it is that I want. I don't want to leave future readers with the impression that this is a technically complex or hard to use box, far from. The technical fundamentals are very sound, the GUI is better than most. Put another way, my dad has and uses one of these on a daily basis and didn't require any help getting things setup.
---- Updated 2014-07-13 ----
Small but important update: About 60d ago I added a UPS (CyberPower CP1000PFCLCD PFC Sinewave UPS 1000VA 600W) to protect the power on this host and I have not had ZFS detect any corrupted bits since then. I suspect that the wall power was not clean and the fluctuating power would periodically result in corrupted bits during writes (suspecting it's the brownout/power fluctuation to the drives that cause the data corruption). Regardless of the cause, ZFS caught and fixed the errors.