I have installed OpenSolaris on a VM in my home network, and updated it to the latest bits (129 at the time of this writing), which supports ZFS’s new block-level deduplication. I’m delighted with the update – it’s reasonably fast (a small office server probably won’t notice the load increase at all) and it does exactly what it’s supposed to – only storing one copy of the blocks in a file, even when the blocks are present in different files. It’s a great way to save space on disk, especially if you are running backups of multiple computers, as there will be significant overlap between them.

This opens the way for me to get serious about rebuilding my two Solaris file servers, as there is no upgrade path from the Nevada builds to the OpenSolaris platform. I’ve already reinstalled the smaller of the two boxes, as it’s only remaining tasks had been running a backup DNS server and processing my network logs. I’m going to have to be much more careful when it comes to moving the larger server, as it hosts all my digital media, including my Aperture photo library. So, it’s time to test extensively on the new box and make sure any initial gotchas are ironed out.

The first interesting test was opening up all of my tar’ed and gz’ed home directory backups and dumping them into a dedup-enabled filesystem on the new system. It looks very promising – it’s a much more efficient way to store files as %90 of the data between backups hasn’t changed. Once I move all the old files off the zfs pool, I’ll be able to report the difference in space, and the reported dedup ratio.