I'm in the LVM-is-better camp, and don't have a whole lot to add given that the major points have been addressed.
But here's one suggestion I've always given to people who work with me: rather than leaving it to hearsay, run your own benchmarks. VMs are cheap, so build two and run your favorite benchmarks to see if you can measure a difference. Don't know of one? Try this one, it's a trusty easy-to-compile standby: http://sysbench.sourceforge.net/ I found that with Virtualbox running under a Linux 2.6 host, LVM was substantially faster than bare metal (hard to explain other than that there is some write-through caching going on at the host O/S level). -rich _______________________________________________ Discuss mailing list Discuss@blu.org http://lists.blu.org/mailman/listinfo/discuss