| |||||||||||||
Btrfs aims for the mainlineBtrfs aims for the mainlinePosted Jan 8, 2009 20:52 UTC (Thu) by masoncl (subscriber, #47138)In reply to: Btrfs aims for the mainline by lmb Parent article: Btrfs aims for the mainline
Andrew Morton recently asked about why the device management in the FS makes sense. The thread covered a bunch of ground:
http://article.gmane.org/gmane.comp.file-systems.btrfs/1764
The most important reason in my mind is connecting the transaction domains of the storage management with the rest of the filesystem. It makes it much more efficient to move storage around without having to wonder if the FS is currently using it.
(Log in to post comments)
Btrfs aims for the mainline Posted Jan 9, 2009 0:29 UTC (Fri) by roblucid (subscriber, #48964) [Link]
Interesting.
Presumbably there's similar issues with the "array of blocks/bytes" idea, when it comes to disks themselves. It shows with Flash, where it'd be nice to let the device know blocks are unused. Perhaps with bad block reallocation to on disks, which might have unpleasant consequences for journal files (without barriers).
It used to be, that disks were always damn near full, so you didn't really gain anything by not doing a raw partition copy when transfering data from one disk to another.
|
|||||||||||||