MAGNETIC DISK STORAGE
Magnetic disks used a similar recording system to the drums, but arranged on magnetic coated platters, made of aluminum, approximately 1ft in diameter. Initially fixed in a stack in a cabinet, the disks later came as a removable 'pack' of 6 platters. These packs could be 'mounted' vertically (spin horizontally) on a drive, as required, and then replaced with another to suit operating requirements.
Read write heads contained in the drive were aligned to move in towards the center of the disks,or and addressed data on both surfaces of the platters (though not the topmost or lowermost surface). A 'Track' is the circular path of one head on one platter. A 'Cylinder' refers to addressing the same track on each platter in the disk pack, forming an imaginary cylinder. The drives spun at speeds in the hundreds of RPM range. The heads were driven by hydraulics, step motor, or voice coil, and had either a servo system or a servo track for address location. The time taken to get the heads on track is the 'seek time'. Capacities ranged from 10 to 100MB as development improved. Much time was spent aligning the heads on track, and recovering from crashes!
The HDA
The next development was the HDA, or Head and Disk Assembly. Essentially similar to the 'disk pack', but fixed in the drive cabinet, the HDA came as a sealed unit with heads ready aligned in the factory, the disks unable to be accessed by the operator (or engineer). The read/write heads were controlled by a voice call mounted in the drive.
The idea was to improve the performance by 'eliminating' the entry of dust to the HDA. Improved capacity was achieved by allowing finer tolerances to be set in the assembly plant. Capacities were improved over the range of 100 to 600MB.
Although there was a big improvement in reliability, there was still deterioration in the magnetic surface. A technique known as 'Bad-Spotting', first developed to extend the life of drums, could be used by the engineer to specify a location on the disk surface which had become unreliable. The failing address or area would have been identified by monitoring the system fault logs, showing addresses which had to be re-attempted.
Bad-Spotting made use of 'spare' areas on the same track as the faulty area. A special type of read would identify the area, and the entire track after this 'bad-spot' would be shifted, so that the bad area was skipped. Alternatively, if no spare area was available, a separate track, specifically reserved for this purpose would be used to relocate this address.
A bad spot table kept a record of all relocated addresses. Following this exercise the disk had to be re-initialised or formatted to use the new physical locations.
Eventually the disk surface would deteriorate to such an extent, or a head crash would occur, requiring the HDA to be replaced.
The PC
The advent of the PC and the requirement for small parts accelerated the development of smaller disks. When the pc disk capacities entered the 100MB and up range, they became viable alternatives to the large HDAs. These drives resembled a miniature version of the disk drive cabinet. They had a stack of several platters and heads arranged on arms. The stepper motors used a worm drive arrangement to locate the correct track.
A cabinet of PC style disk drives might contain several banks of 8 drives to a bank, giving better total capacity than the HDA, and possibilities of redundancy within the cabinet(see below). Also air conditioning requirements were relaxed, and a smaller footprint (space taken by the cabinet) was possible. As disk drives improved they were incorporated into the cabinets in ever increasing capacity.
Redundancy became possible relatively cheaply with the lower cost of drives. The principle is to write identical data to two drives for each address, so that, if one drive fails, the other can be used. A sophisticated version would switch in a spare drive to take over from the defective one, allowing the faulty drive to be replaced, while the user continues on without interruption.
The RAID system (Redundant Array of Independent Devices) is a refinement of this technique. Redundancy can be had in pairs of drives, as above, or in stripes across a series of drives, or a combination of both.
In part 3 we look at magnetic tapes.
Tony is an experienced computer engineer. He is currently webmaster and contributor to http://www.what-why-wisdom.com. A set of diagrams accompanying these articles may be seen at http://www.what-why-wisdom.com/history-of-the-computer-0.html. RSS feed also available - use http://www.what-why-wisdom.com/Educational.xml |
No comments:
Post a Comment