This article discusses the fundamentals of disk drive technology.
Disk Geometry
Today's disk drives are based on magnetic recording principles. Data is stored on disks by magnetizing particles, which will ultimately correspond to the computer's data of binary ones and zeros. The disk drive itself contains a series of disks, which are termed platters. Each platter has two recording surfaces - top and bottom. These platters are stacked on top of one another in the same vertical plane with sufficient space between them to accommodate recording heads. The heads are responsible for reading and writing of data. Reading is the process of retrieving data from a disk drive, and writing is the process of storing data on a disk drive. The platters are usually made up of a lightweight material such as smooth glass or a light metal and coated with a substance that can be magnetized.
Data is stored on each surface in a series of invisible concentric rings, which are termed tracks. If each track is viewed within the same vertical plane, we get a structure that resembles a cylinder. The number of cylinders on a disk drive is the same as the number of tracks per recording surface. Each track is further divided up into sectors.
Capacity and recording
A sector is further subdivided into a collection of bytes. Each byte is made up of 8 bits of information. Most Operating Systems have abstracted the three dimensional disk geometry into a series of logical blocks, where the total blocks on a disk is equal to the number of sectors multiplied by the number of data heads multiplied by the number of cylinders. This technique is known as Logical Block Addressing (LBA). Most blocks are 512 bytes in size. The capacity of a disk drive can be calculated by multiplying the number of sectors by the number of data heads by the number of cylinders by the number of bytes per sector, or by multiplying the number of blocks by 512. In the disk drive environment a Megabyte is equal to one million bytes rather than the binary Megabyte of 2**20 (1048576) bytes
The platters spin at a constant speed and the recording heads move laterally across the tracks. The heads do not come in contact with the surface of the disk but instead they ride on a cushion of air.
The closer the heads come to the surface of the disk the denser the recording becomes. The distance between the head and the surface is extremely close and is achieved by extremely precise mechanical tolerances. An instance of a head coming into physical contact with the recording portion of a disk is termed a head crash and usually has fatal consequences. When the disk is spun down the heads will retract to a specially lubricated portion of the surface which is termed the landing zone. Some disks when manufactured are not 100% error free but have a few minor blemishes. These blemishes are known as bad areas of the disk and are recorded on the disk in an area known as the Manufacturer's Defect Area.
More recent developments have introduced perpendicular recording where the magnetization takes place vertically rather than horizontally across the disk surface. This has led to greater bit densities.
Access times
Initially drives must be formatted before use. This process is analogous to preparing a blank set of pages for writing by ruling out lines and margins as well as the insertion of headers and footers. This extra information, which is held within each block, will actually take up some storage space, so the formatted capacity of a disk drive is less than the unformatted capacity. This distinction is very important to the user who is only concerned with usable capacity.
Because accurate retrieval of data is crucial for the end user, a method of incorporating error detection and correction is incorporated within each sector. A common method for detecting errors is a scheme known as Cyclic Redundancy Checking or CRC. A more advanced technique is to employ an Error Correcting Code, which contains redundant information to correct data errors. Some advanced forms of disk software will monitor this correction threshold and, when it exceeds a pre-determined limit, will automatically retire the defective block and transparently redirect the data to a good area of the disk.
The access time of a disk is the time taken to retrieve the data. This access time is made up primarily of two factors, one factor is the rotational latency and the other is the seek time. Rotational latency is the time taken for the desired sector to pass under the recording head. SATA disks typically spin at a rotational speed of 7200 R.P.M. so this means that it will take 1/60th of a second to make one revolution (1/60th of a second is equal to 8.3 milliseconds). When calculating the rotational delay we calculate the average delay, which is the time taken to perform one half of a disk revolution = 4.17 milliseconds.
The seek time is the time taken to move the heads to the desired track and this tends to vary from drive to drive but an average seek time of 10 milliseconds is not uncommon. When we add the rotational latency and the seek time together we get the access time. In this example, the access time would be quoted as 14 milliseconds. The time taken to actually transfer the data is known as the data transfer rate, but as this is normally very fast compared to the latency and seek times, we can usually ignore it. Modern disk drives have rotational speeds of up to 15,000 R.P.M., which result in very fast access times.