OpenZFS on OS X1.5.2

I would like to share my thoughts with others and welcome feedback. I have used ZFS before it appeard on OSX and used it on OSX since 32bit PPC days... The first thing to keep in mind, is that there's been several attempts to kickstart this filesystem on OSX and even Apple intended to use it before Oracle bought out Sun Microsystems. These attempts are not a reflection of ZFS's features itself, but more so an issue of execution on the Mac OS X platform. Articles exist on the web that suggest once you starting reading around 12TB of data on spindles, you're statistically guaranteed to have encrounted a read error. (http://www.zdnet.com/blog/storage/why-raid-5-stops-working-in-2009/162). This is also known as silent data corruption, bit creep etc etc. The merit of the article or how accurate the "12TB" value actually is, isn't really something that should be discussed here, but the message is clear – the larger hard-disks become, the more likelihood read errors will occur when you read/copy the data. Think about his for a second: How often do you replace your storage? let's conservatively say every 5-7 years. (some only do it if a disk fails) And how long do you propose to live? let's say the "y" generation is going to reach 80+ no probs. So, if you transfer your data every 5-7 years, that's 12 occasions (minimum) of migrating/copying your data throughout your life. How confident are you that all your data will still be readable when you want to hand down photos of loved ones or movies of your children? If you're not storing your data that can be guaranteed consistent-on-disk by using block checksums, you may want to rethink a few things. Also, what I'm talking about here has nothing to do with backup strategies – that's another topic for discussion another time, but realise that if you cannot guarantee consistency-on-disk, all you're doing is potentially moving a hidden problem to your backup strategy. This all sounds fairly drastic and harsh, and this issue is essentially masked by current disk sizes, but with 6TB drives available, it's something you might want to carefully consider for the future. Being able to use checksums to verify your data is just one of the many features of ZFS, there are many more that no other filesystem can offer natively on OS X and so it's for this alone that I rate ZFS so highly. HFS is should've been retired years ago, (http://arstechnica.com/apple/2011/07/mac-os-x-10-7/12/) Even Apple know this by there decision to adopt ZFS back when Mac OS X 10.5 was being developed. How do you feel about your data sitting on HFS now? Hat's off to the guys developing ZFS on OSX, you guys rock! Regards, R.

v1.6.1 is out and still compatible from 10.8 to 10.12.
(I've tested it on 10.9 and all seems ok)

It doesn't take long to "feel" the superiority of ZFS over HFS: a few speed tests on copying huge files reveal faster writing and, if you activate compression, a space gain (a 37 Gb HFS became 35.4 Gb ZFS!) — it's actually even faster with compression on! Of course, the scope of ZFS is rather on data security, but I can only rely on what I've read, since it's just been 2 weeks I'm testing it (indeed because I discovered hundreds of bad blocks on an HFS drive... ;-) ). The principles behind ZFS are pretty reassuring: "copy-on-write", self-repairing mirrors... And the OpenZFS dev team is really responsive! I only give 4.5 stars, just because it's not fully integrated in OSX (no GUI, DiskUtility.app and other utilities don't deal with it directly and you will have to use Terminal for setting up and managing "pools"). Yet, it's very easy to create simple pools, and once you understand the concepts of it, you easily create a ZFS partition on your HFS drive without any issue. If you care about your data, ZFS is mandatory.

I have been running this since the day Mavericks came out on a production system (before that on a virtual machine for test purposes). At this point I'd recommend it to anyone who wants to benefit from the advantages of running zfs. It is rock solid, and has all the niceties that one would expect (installer, disks automatically mount). It provides "zvols", which means you can have a zfs block device with all the zfs fun stuff (snapshots, auto healing, ...), and format it as HFS+. I'm using that to have TimeMachine backup to zfs.

The developer's website states "It's compatible with OS X 10.6 through OS X 10.10 (Snow Leopard, Lion, Mountain Lion, Mavericks, and Yosemite). " Why is this marked here as needing OS X 10.8 and above?

I would like to share my thoughts with others and welcome feedback. I have used ZFS before it appeard on OSX and used it on OSX since 32bit PPC days... The first thing to keep in mind, is that there's been several attempts to kickstart this filesystem on OSX and even Apple intended to use it before Oracle bought out Sun Microsystems. These attempts are not a reflection of ZFS's features itself, but more so an issue of execution on the Mac OS X platform. Articles exist on the web that suggest once you starting reading around 12TB of data on spindles, you're statistically guaranteed to have encrounted a read error. (http://www.zdnet.com/blog/storage/why-raid-5-stops-working-in-2009/162). This is also known as silent data corruption, bit creep etc etc. The merit of the article or how accurate the "12TB" value actually is, isn't really something that should be discussed here, but the message is clear – the larger hard-disks become, the more likelihood read errors will occur when you read/copy the data. Think about his for a second: How often do you replace your storage? let's conservatively say every 5-7 years. (some only do it if a disk fails) And how long do you propose to live? let's say the "y" generation is going to reach 80+ no probs. So, if you transfer your data every 5-7 years, that's 12 occasions (minimum) of migrating/copying your data throughout your life. How confident are you that all your data will still be readable when you want to hand down photos of loved ones or movies of your children? If you're not storing your data that can be guaranteed consistent-on-disk by using block checksums, you may want to rethink a few things. Also, what I'm talking about here has nothing to do with backup strategies – that's another topic for discussion another time, but realise that if you cannot guarantee consistency-on-disk, all you're doing is potentially moving a hidden problem to your backup strategy. This all sounds fairly drastic and harsh, and this issue is essentially masked by current disk sizes, but with 6TB drives available, it's something you might want to carefully consider for the future. Being able to use checksums to verify your data is just one of the many features of ZFS, there are many more that no other filesystem can offer natively on OS X and so it's for this alone that I rate ZFS so highly. HFS is should've been retired years ago, (http://arstechnica.com/apple/2011/07/mac-os-x-10-7/12/) Even Apple know this by there decision to adopt ZFS back when Mac OS X 10.5 was being developed. How do you feel about your data sitting on HFS now? Hat's off to the guys developing ZFS on OSX, you guys rock! Regards, R.