So the code for ZFS has been released and is "Open", but not "Free" (its
licensed under the SCDL).
For those of you wondering what ZFS is, here's a collection of links.
Its all pro-Sun, of course.
http://blogs.sun.com/roller/page/bmc?entry=welcome_to_zfs
A port to linux is "being investigated":
http://www.sun.com/emrkt/campaign_docs/expertexchange/knowledge/solaris_zfs_gen.html#10
ZFS was primarily designed for deployment in an enterprise environment
where performance and scalability are extremely critical. It was
engineered with a totally different set of assumptions focused on
managing a huge amount of storage, so comparing it to other contemporary
filesystems like Reiser would probably not be a good use of time.
That said, a lot of the 'features' found in ZFS are not innovative
per-se, and the kernel-userland interface is a bit gross (but who reads
the source these days?) Not sure how to define a good interface to LVM
for dynamically sizing partitions ... no problem, just hack the entire
LVM into the FS. Not sure how to put checksumming in the block layer to
give reliable storage ... no problem, just hack parts of the block layer
into the FS. Parts of ZFS appear to be a huge layering hack to me, but
as I said, who reads the source these days?
For the true hardcore linux user, most of what ZFS and Solaris provide
in terms of innovation can be replicated in Linux with a bit of
fiddling: a lot of what ZFS does can be done with md and lvm, zones are
like vserver (plus Linux also has UML and Xen, though Xen should be
coming to Solaris too).
Still, ZFS is a package that is advertised to work well. With its
self-healing capabilities and support for a new high-reliability data
replication model called RAID-Z, a Linux port of ZFS could potentially
become a very popular choice for commodity file servers in Enterprise
environments, but the benefits of ZFS on a desktop Linux system are
debatable.
