On Sep 11, 2012, at 2:57 PM, Rodger Donaldson wrote: > You should probably talk to the client about the use cases where they > see themselves as getting something out of btrfs that they won't get out > of ext4+LVM or XFS+LVM.
I realize this adds nothing to the technical discussion, but anecdotally: my team rode XFS+LVM to big wins in a number of areas. Off the top of my head... 1) the "thundering herd" problem of all linux servers simultaneously trying to check their ext3 filesystems after IPLing CP 2) immaturity of tools for online resizing of ext3 at that time, where they would or would not work apparently at random 3) forcing us to improve our processes to no longer rely on access to the E2CMD MODULE during "tricky" maintenance activities (and DR testing), which interacted badly with ext3 log replays on next reboot and caused pretty predictable data loss if it was used on a filesystem that had not been shut down cleanly. We also saw modestly improved linux I/O rates at somewhat lower CPU cost, compared to ext3. TBF ext4 wasn't available at that time. I can't comment on its characteristics. We did have to keep closer track of UUIDs (both for LVM and XFS) to keep things sane in the event of attaching one guest's mdisks to another guest. But this was done as two or three additional minor steps in our cloning automation, so in practice it was nothing. ok r. ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For more information on Linux on System z, visit http://wiki.linuxvm.org/
