Data structure in disk.

Disk Structure and Partitions

A hard disk is physically composed of a series of flat, magnetically coated platters stacked

on a spindle. The spindle turns while the heads move between the platters, in tandem, radially

reading/writing data onto the platters.

Fig. 3. Physical Disk Structure

Disk tracks, cylinders, and sectors

A disk is divided into tracks, cylinders, and sectors. A track is that portion of a disk which

passes under a single stationary head during a disk rotation, a ring 1 bit wide. A cylinder is

comprised of the set of tracks described by all the heads (on separate platters) at a single seek

position. Each cylinder is equidistant from the center of the disk. A track is divided into segments of

sectors, which is the basic unit of storage.

On Sun systems a sector is 512 bytes (1 disk block) of data, with header and trailer

information. The latter make it possible for the controller to identify sectors, detect data errors, and

perform error corrections when necessary. The actual layout of a disk sector will vary depending on

the controller, but should look something like that shown in Fig. 2.3. There are two Preambles and a

Postamble (whose sizes may vary due to rotational speed, etc., and are disk dependent). The Header

field lets the controller know where the head is positioned, and the ECC field is for error correction.

Fig. 2. Disk Platter

The number of sectors per track varies with the radius of the track on the platter. The

outermost tracks is larger and can hold more sectors than the inner ones. These outer tracks also

spin faster under the head than do the inner ones, because while the angular speed remains the

same, the larger circumference results in more sectors spinning by in the same period for the outer

tracks. Disk blocks are numbered starting at the outermost track, so put the data you expect to

access most often on partition, or slice, 0.

Cylinder group

SunOS uses the Berkeley fast file system which uses cylinder groups. A group is formed

form 32 or fewer cylinders on a disk (default 16). Each cylinder group has a redundant copy of the

superblock, space for inodes, list of available blocks, and a list of data block usage within the

cylinder group. Data blocks are spaced to minimize rotational delays and to keep blocks of the same

file close together. By grouping cylinders in this way we reduce the amount of head movement, on

average, required to access a file. The inode describing the file, and the data for the file, are likely to

be in the same physical area of the disk. The position of the redundant superblock within each

cylinder group is varied, so that they don't all reside on the same disk platter. This helps to insure

that you can recover in the event of the loss of the primary superblock.

Physical disk structure

Disk formatted in MS-DOS has physical structure (Fig. 5).

1. Boot record has special program for checking is disk system or not. Checking criterion is

very simple: first two files in a root directory must be IO.SYS and MSDOS.SYS. These two

files must be in a root directory exactly disposed in this order. If disk is system boot record

every time is disposed in first sector of zero track of zero cylinder. After physical disk sector

destruction is no more in use. One physical disk division into independent logical parts for

fixed disks is called section. On of the section is used for MS-DOS booting. This is an active

section. Information about executed disk division is placed in special called Partition table.

This is a part of Master Boot Record. This record places in the first sector of the zero

cylinders of the zero tracks. When recorded program got control, it defines which one is

active and from which cylinder this division begins.

2. First FAT copy (read about File Allocation Table)

3. Second FAT copy. FAT is very important supporting data integrity. That’s why MS-DOS

has two identical tables’ copies that give opportunity “save” contents on disk after damaging

using the copy for the file access.

4. Data domain. The ret space after enumeration is available for the file location. First cluster

in data domain every time is cluster number two. This does not mean that all enumerated

fields are in two clusters. First two FAT elements are used as disk format indicators.