© Tony Lawrence, aplawrence.com
Sun's new file system for Solaris 10. Apparently unrelated to IBM's zFS
What's so impressive? Well, let's start wth that it's a 128 bit file system: http://blogs.oracle.com/bonwick/date/20040925#128_bit_storage_are_you notes that
.. fully populating a 128-bit storage pool would, literally, require more energy than boiling the oceans.
That pretty well covers it, I guess.
ZFS is supposed to be easier to comprehend and administer. I'm all for that: I have often complained that filesystems like Veritas (a former favorite of Sun) can be very difficult. I have yet to put up a Solaris 10 box so am not ready to agree that it is easy yet, but any improvement will be welcome.
Every bit of data is protected by a 64 bit checksum. That ought to be impressive. The Sun folk also do a lot of talking about how its copy-on-write design eliminates the need for fsck:
Data can be corrupted in a number of ways, such as a system error or an unexpected power outage, but ZFS removes this fear of the unknown. ZFS prevents data corruption by keeping data self-consistent at all times. All operations are transactional. This not only maintains consistency but also removes almost all of the constraints on I/O order and allows changes to succeed or fail as a whole. All operations are also copy-on-write. Live data is never overwritten. ZFS writes data to a new block before changing the data pointers and committing the write.
(from pan style="text-decoration:line-through;color:red"> http://www.sun.com/2004-0914/feature/(link dead, sorry))
Ok, so it's transaction based. That doesn't entirely eliminate the need for fsck-like integrity checking, but it does make it faster.
ZFS is also "self-healing":
As part of Sun's quest to build truly self-healing systems (see the September 7 Sun.com feature), ZFS can self-heal data in a mirrored or RAID configuration. When one copy is damaged, ZFS detects it via the checksum and uses another copy to repair it. No competing product can do this. Traditional mirrors can only handle total failure of a device. They don't have checksums, so they have no idea when a device returns bad data. So even though mirrors replicate data, they have no way to take advantage of it. By contrast, the end-to-end checksums in ZFS allow it to find and fix bad blocks--with nineteen nines certainty--automatically.
Well, it's all very interesting. Snapshots, of course, and tantalizing other things like "Multiple block sizes, automatically chosen to match workload", which certainly would be a neat trick. Apparently you can choose block size on a file-by-file basis.
I'll be interested to learn more about this. There are, of course, already nay-sayers who say ZFS will be too slow. Unless it is dog-slow, the other benefits could easily outweigh that, and we need to remember that hardware gets faster every year. If improved semantics for software require faster hardware, the hardware will catch up. That's always been true.