On Sun, Feb 13, 2011 at 7:07 AM, Jean-Christophe Deschamps <j...@antichoc.net > wrote:
> > >So if the drive is 5400 rpm, 227 is much more than 5400/60=90 and even if > >it's 7200 (manufacturers sometimes upgrade drives inside portable hd > >without > >prior notice), it's still twice as much as 7200/60=120. > > 5400/60, 7200/60 ... those values rely on the assumption that > successive LBAs are mapped to successive physical sectors (512 or 4K, > whatever) on the same face of the same plater. Is it obvious that all > today's typical stock drives actually implement only that simple old > scheme and not an untold mix of various interleaving techniques? > This is true for the case of writing multiple records or multiple sectors. For example, if you have a drive with 5000 sectors per track and you write sector 1 of the track with a hard sync, you may have time to write sector 2500 with a hard sync in the same revolution. Or maybe you can write every 500 sectors with a hard sync in the same revolution, giving you a commit rate of 10 per revolution or 900 commits per second on a 5400 rpm drive. But what I postulate is that you can't physically write *the same* record over and over more than 90 times per second on a 5400 rpm drive, unless the drive, OS, or filesystem implements something like wear-leveling, where the physical location of sectors is constantly changing. Jim -- HashBackup: easy onsite and offsite Unix backup http://www.hashbackup.com _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
