Home

Hardware

HDD3: Data storage structure (disk's geometry), Coding methods

Physical structure (Geometry)

• It is possible to define the exact data position on the disk by help of the physical structure of the hard disk. 

• Number of heads - When configuring, you can meet the setup of the heads where you have to enter the number of used surfaces. The rule of two heads for a platter is not always valid. To be compatible with first disks the value usually do not exceeds 16 heads. (in the BIOS system options - max. 255)

• Track - Each disk has several thousands of tracks. Each track is formed by one of many coaxial circles. Different size of the tracks in the middle and at the edge of the disk is a kind of particularity and also a problem. Particular tracks are numbered from 0 to higher numbers (from the outer to the inner track). There is the FAT table (File Allocation Table) placed on the track 0. In the BIOS system the number of tracks is represented by help of cylinders. 

• Cylinder - the tracks identically placed one over another in particular platters mean a cylinder. Heads and cylinders are placed one over another thanks identical geometry and servo mechanism (there must be defined the number of the surface on which the head records and reads) (in the BIOS system options - max. 65536)

• Sector - the least unit on the disk having the typical size of 512 bytes. To define the size of the disk there is given the number of sectors for a track. This data is not fully true (each track is different!), the value is represented by an average and defines rather the logical configuration (usually 63). (in the BIOS system options - max. 255 - in some only 63). According to the size of the section, the operating system arranges several sectors to one cluster. So, the operating system more effectively works with more sectors at a time and instead of sectors it refers to clusters. About the cluster size under Windows 2000 you can read in the article called "File systems not only under Windows 2000".

• Why do I write about the logical structure? In older disks it was necessary to enter the values of particular data to the CMOS memory. As for the geometry, the disk is uniform, but in fact the tracks are different and the number of sectors for a track is also not identical. In present types of disks and computers, it is not necessary to enter the values manually. By help of BIOS settings the data are recognized automatically.

• Other data entered to the CMOS memory (most of the present disks do not require these data): 

• Write Precompensation Cylinder - solving the problem of defining the sectors on the disk having different number of tracks on the outer and on the inner part of the disk. (The cylinder number that at which the disk drive changes the write current).

• Landing Zone - a sector for parking the heads when the disk is off. This was important for older disks with a stepper. In present disks the heads are parked automatically onto the beginning of the disk.

Logical data structure

• Considering many limitations and the wrong reasoning about the identical number of sectors on a track, there comes the logical structure to overcome the limitations and for the right value interpretation.

• ZBR (Zone Bit Recording) - recording method using different number of sectors for a track (the sizes of the sectors are identical, but the number of sectors is different according to the position of the track on the disk). On the inner part of the disk the track is longer, so more sectors can be placed in there then in some on the outer part. In older disks, there was the same number of sectors in all tracks. 

• The logical structure makes the disk's structure easier - it simulates identical number of sectors for a track on the inner and on the outer part of the disk. Thanks coming ZBR it is not recommended to try the low-level formatting on the disks. 

• There is also another type of the logical structure. Considering some limitations of the BIOS system, some data are translated to a logical organization - e. g. in a disk having higher amount of cylinders the number of heads increases and the number of cylinders lessens to a reasonable value, see BIOS. Sometimes, instead of logical addressing there is also the sector coding term used. Recently, this is replaced by the LBA addressing. Sector coding is used in disks where the parameters do not correspond with some limitations. The solution is a mathematical change (e. g. instead of 32000 cylinders and 2 heads we can use values of 4000 cylinders and 16 heads when addressing)

 Organization of recording and reading - Coding methods

• Data recording - the heads are moving over the surface and produce the magnetic field of different polarity (the polarity is created by changes of the direction of the electric current; so, the principle of the heads is an electromagnet changing its polarity very quickly). 

• Data reading - the heads are moving over the disk's surface and the changing polarity indicates the electric current in the heads, which is then translated to the clear data definition. 

Particular methods

• The recognition of particular bits depends on the recognition of the magnetized and not magnetized places. The problem is in distinguishing several bits (0 or 1) going one after another, where the places having the same polarity are very difficult to be defined. The solution is to use coding methods using different time information, distinguishing symbols placed on the disk and other ways. The principle of modernization of particular method is using of the least number of distinguishing symbols, lowering the size of the point for bits recording with the result of saving space for the data itself and increasing the data record density. In the most modern methods it is possible to distinguish several identical bits lying one next the other. Method from the oldest to the newest:

• Frequency Modulation (FM) - method used in first types of disks with ST506, ESDI interface. There is a distinguishing symbol between each bit.

• Modified Frequency Modulation (MFM) - improved method of FM coding. At present, you can meet this in diskettes. Over against the FM method, this one is able to double the data density thanks reducing the number of distinguishing symbols.

• Run Lenght Limited (RLL)

• Partial Response Maximum Likehood (PRML)

• Extended PRLM (EPRLM)

• Extended Partial Response (EPR)

Related Links

Web links»Hardware»Storage

Print Print article | Recommend Send link