On Fri, 26 Jun 2009, Matt Kivela wrote:
If someone can point me towards some references to research that say ZFS and OpenSolaris on x86_64 can do > 4k block sizes, and there's nothing in my hardware configuration that screams "can't do it!", then I can get my ducks lined up and ask the boss for a few days of lab time to play with OpenSolaris.
Presumably by now you have found the abundant research. Solaris in general does not seem to couple the MMU page size with the filesystem block size. Since at least Solaris 9 it has been possible to use larger MMU page sizes, and this does not require a change to existing filesystems. The available MMU page sizes are architecture specific, but Solaris does not base its filesystem access on MMU page size for either SPARC, x86, or AMD64.
ZFS does indeed write data using larger blocks, and at 128K per block by default.
Bob -- Bob Friesenhahn [email protected], http://www.simplesystems.org/users/bfriesen/ GraphicsMagick Maintainer, http://www.GraphicsMagick.org/ _______________________________________________ storage-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/storage-discuss
