|
From: dmccunney <den...@gm...> - 2012-02-08 21:48:39
|
On Wed, Feb 8, 2012 at 4:10 PM, Bertho Grandpied <y31...@ya...> wrote: dmccunney <dennis***> wrote : > >> (albeit much of the additional space is lost slack at the user files level ...) > > Slack space will be a real issue? I don't see how. > > There would be a net loss for partitions sizes and file systems comprised of less than > 4K bytes per cluster. You are right, not a big problem, in general. I was repeating what > I read, without too much thinking. Slack space was an issue under FAT12/16, because the minimum unit of disk MS-DOS would read/write in one operation was the cluster, and there was a hard limit to the number of clusters possible. (65,536 under FAT16.) Cluster size varied depending on volume size, with a 2GB volume and a 32KB cluster being the top end. Only one file could be on a cluster, so it was possible to have a one byte file that would require 32KB on disk. Disks were smaller and more expensive, and various schemes were used to minimize slack space to store the most data possible. FAT32 removed the limit on the number of clusters, but the concern wasn't slack space: it was allowing larger logical volumes on drives. Partitioning to create multiple 2GB volumes seen as separate drives got old fast. Current hard drives are huge and cheap, and getting bigger and cheaper. I wouldn't call slack space a problem in particular there days, let alone in general. The amount of storage lost to slack apace simply isn't big enough to be worth addressing. > Czerno ______ Dennis |