Part 4: Data Storage continued...

Chapter 11: How Disk Drives Save Information

The capacity, size and performance of disk drives have changed dramatically since their introduction as standard equipment on the first IBM XT in the 1980s.

How a Floppy Drive Stores a Little  

• ·         When a 3.5” disk is pushed into the drive levers pull back the shutter to expose the cookie
• ·         Two read/write heads move until the barely touch the cookie on either side and generate magnetic pulses to change the polarity of metallic particles in the disk’s coating, a motor spins a shaft which rotate the disk from the hub while a stepper motor positions the head.
• ·         The drive’s circuit board receives signals that control the head, if a write command is received an LED/sensor combination check to make sure the write protect tab is not open (meaning the disk is protected) before commencing the write

How a Hard Drive Stores a Lot

• ·         A sealed metal housing protects the drive from dust which could cause a crash due to tolerances less than the thickness of a human hair between the read/write heads and the disk.
• ·         Commands are received from BIOS through the disk controller which converts generic commands into ones that are specific to that drive.
• ·         A spindle spins as many as eight coated platters.
• ·         When reading or writing the OS first checks a file allocation table or master file table to determine the physical location(s) of the desired file or to find an available cluster for writing and passes this information on to the drive’s built-in disk controller which in turn instructs the head actuator on how to move the read/write arms to align the heads to the correct cluster. If writing it records the location of the file’s clusters in the FAT or MFT.
• ·         Typically there is a gap of only 2/10e6 inch between the heads and the disk surface.

How a Mirrored Drive Array Protects Files

• ·         RAID – redundant array of independent (or inexpensive) drives can speed performance and provide second-to-second protection against disk crash depending on how the array is configured.
• ·         When data integrity is more important than speed and there are only two hard drives, the best solution is a mirrored disk drive (or RAID 1). The RAID controller replaces an ordinary disk controller with software to handle arrays, writes every file to two or more drives simultaneously. In case of a read error caused by surface defect or crash, the controller reads the intact version of the file from an undamaged drive.
• ·         To read a file from a mirrored array, the controller reads alternate file clusters from each of the drives and pieces them together for delivery. This makes reads faster proportionate the the number of drives in the array (2 = 1/2, 3 = 1/3 the read time).

How Striped Drive Array Boosts Performance

• ·         For faster reads and writes a RAID is set up as a striped array (RAID 0) of two or more hard disks. The RAID controller divided files into several pieces and writes different sections to different drives at the same time. To read the controller simultaneously pulls each part of the file from the different drives on which it’s stored then pieces the parts of the file together in correct order when it writes them to RAM.
• ·         In a striped array with parity (RAID 4), the controller writes striped data to all but one of the drives which is used for error checking.  The controller’s array software performs a Boolean XOR operation on the data written to the drives and writes the result (a parity bit) to the remaining drive. An XOR operation results in a 0 bit whenever two like bits are compared and a 1 bit whenever two dissimilar bits are compared. If more than three drives are in the array, the first two are XORed and the result is XORed with the next drive, etc., until all the data drives have been XORed and the final result is written to the parity drive.  Later, if one of the files is damaged or a drive crashes, the controller can deduce whether the missing bits are 1s or 0s. This reverse XOR can also be used to repair minor media defect damage on the fly.

Chapter 12: How Little Things Make Drives Faster and Bigger

Scientific progress is incremental, most advances are made by plodding toward incremental gains rather than by momentous discoveries. Computer speed has always been limited by computer storage speed.

How Small Changes in Drives Get Big Results

• ·         The ZIP drive increased data volume and access speed over floppy storage, but allowed for portability.
• ·         A smaller head allowed 2118 tracks per inch, compared to 135 tracks per inch on a floppy disk.
• ·         Additional storage capacity came from coating the cookie with the same magnetic particles as a S-VHS video tape which have a higher energy level and therefore not as easily magnetized. So the magnetic field from the write head affects a smaller surface area.
• ·         Change from radial sectors (which waste storage space on the outer tracks) to zone recording (also used for hard drives) so that the same recording density is used throughout the disk.
• ·         Perpendicular recording

o   The superparamagnetic effect limited storage capacity because proximity caused charges to interfere with one another.

o   Iron core redesigned to ride close to the disk’s surface and force a stronger magnetic field through a smaller area leaving particles magnetized perpendicular to the path of the magnet.

How File Compression Makes Files Smaller

• ·         Makes files smaller by eliminating redundancies in the file (lossless compression)
• ·         Uses some variation of LZ (Lempel and Ziv) adaptive dictionary-based algorithm, which examines the file for recurring patterns of data which are written to a dictionary stored as part of the compressed file. The compressed file is stores as a reduced version with pointers indicating where the omitted dictionary sections can be found. Adaptive means that the algorithm constantly looks for more efficient data patterns and changes entries on the fly.
• ·         The type of file determines how much compression takes place, length is also a determinant in degree to which a file can be compressed. On average files can be compressed 50%, but some can be compressed to only 20% of their original size.
• ·         Graphics files often use lossy compression which reduces file size by discarding data such as small variations in color. These files can never be returned to their original state from the compressed version. Utility programs can further compress files, they tend to emphasize storage over speed.

How Disk Defragmentation Works

• ·         Data is written to new disks in contiguous clusters. However as files are deleted, available clusters may not be adjacent. Defragmenting reorders data so that data from a file is in a contiguous set of clusters. Defragmentation helps to speed the read and write actions by reducing the amount of time moving from track to track passing over unrelated sectors.
• ·         Defragging or disk optimization is a software operation that moves the scattered parts of files so that they are contiguous on the disk by first moving out fragmented block to make space and then moving the file data back into the contiguous cluster group. Windows Vista automatically defrags drives when they are idle.