On Wed, Feb 19, 2014 at 01:52:08PM -0800, Andrew Hume wrote: > an almost counter-intuitive finding.
RAID will be a necessary part of the future, but not all of it. Forgive the scattered thoughts; it's been a long day for me. I think that the ZFS and btrfs implementations point the way: checksumming (OK, hashing) all your blocks is necessary to be sure that all the parts of your RAID are robust. In my experience, data is rarely corrupted while on disk - but it becomes vulnerable every time it is moved. Consider desirable features for a successor filesystem to ZFS, btrfs, LUKS and git: stability, resilience, recovery, snapshotting, versioning, encryption, archiving, clustering and network-awareness. Deduplication probably falls out of that for free. Performance will require caching. Multiuser encryption strongly suggests that identity (ownership) will be even more firmly tied to the filesystem. An awful lot of workflows really need versioning rather than snapshotting. It's nice to recover to yesterday's stopping point, but retaining milestones has more meaning to users day in and day out. -dsr- _______________________________________________ Tech mailing list Tech@lists.lopsa.org https://lists.lopsa.org/cgi-bin/mailman/listinfo/tech This list provided by the League of Professional System Administrators http://lopsa.org/
