Section 4 - Storage Systems

Storage Systems

     Four types of hard disk technologies have been used in PCs over the years: the ST506, ESDI, IDE, and SCSI. The first two are no longer in use as the AT computer in which they were used. Most of the PCs in use today use either IDE or SCSI hard disk drives.

IDE Technology

     Integraded Drive Electronics or IDE technology was orginally developed as an alternative to the expensive SCSI technology by Compaq and Western Digital. This technology integrates the controller board into the disk drive.

     IDE, also known as AT Attachment (ATA) interface only needs a passthrough board to connect the device to the motherboard. Some newer designs integrate an IDE controller into the systemboard. IDE supports up to two 528MB drives. A newer version EIDE (Enhanced IDE) enlarges capacity into four multigigabyte drives. Beyond that capacity, you must use SCSI or some other technology.

     IDE driver are low-level formatted at the factory. A low-level format is one that scans the disk storage media for defects and sets aside sectors with defects so that they are not used for data. IDE drives should never be low-level formatted by a user or a technician.

Tape Drives

     Tape drives are used primarily bor backing up large databases or hard disks. Because of its streaming capabilities, it is perfect for recording data in a serial format.

     You can attach a tape drive to the PC as an internal or external device. Tape cartridges come with either 4mm or 8mm tape. The DAT (Digital Audio Tape) is a high-capacity 8mm technology.

CD-ROM Technologies

     The CD-ROM is one of the optical technologies that uses a laser to read its data from its media. A CD-ROM can hold 650MB of data. Data on a CD-ROM is recorded in one long continuous strand beginning on the inside edge. The CD-ROM filesystem is called ISO-9660.

     Most external CD-ROM drives use a SCSI interface and come with their own host adapter cards. Not all SCSI CD-ROMs come with an adapter though. To operate an CD-ROM in the MS-DOS environment, you must load it's device driver with the CONFIG.SYS file. The device driver must then be used by the MSCDEX.EXE application to be assigned a drive letter.

     When installing an IDE CD-ROM drive to a system that has an IDE hard disk installed, you must configure the CD-ROM drive as a slave drive. Some CD-ROM drives only run as slave drives.

Optical Disk Drives

     Like the CD-ROM, the optical disk uses lasers to read and write data. The optical disk can be written to and read from more than once. Optical disks can store up to 6GB of data.

     Optical disks are available in two technologies:

WORM (Write-Once, Read-Many) - An optical disk that can be written to, but only once. The WORM disk drive records data by making permanent physical marks on the media's surface with a low-power laser.
EO (Erasable Optical) - Disks that are erasable, can be overwritten with new data, and can be used much like magnetic disks. The laser erases any existing marks on the media's surface by relaxing the mark, allowing it to be reused.

SCSI Technology

     The Small Computer Systems Interface (SCSI) is a collection of interface standards that covers a wide range of peripheral devices, including hard disks, tape drives, optical drives, CD-ROMs, and disk arrays (RAID). Up to eight SCSI devices can connect to a single SCSI controller by sharing the common interface, called a SCSI bus or SCSI chain.

     SCSI devices are like IDE/EIDE devices in that the device controllers are built into the device itself. Each device on a SCSI bus is assigned a unique device number. These numbers are configured to the device with jumpers or DIP switches located on the device. When the SCSI controller (which counts as one of the eight devices) wishes to communicate with one of the devices on the bus, it send a message encoded with the unit's device number. The device will reply with its number to ensure the proper device. The bus is terminated at each end with a terminating resistor pack.

     The original SCSI interface is now called SCSI-1. They have a 5MB transfer rate, use either a Centronics 50-pin or a DB-25 connector, and have an 8-bit bus. The SCSI-2, also called SCSI Fast-Wide, includes a 16-bit bus (Wide SCSI), and twice as fast transfer rate (Fast SCSI). SCSI-3 include Ultra SCSI, Wide Ultra SCSI, SCSI Parallel Interface (SPI), and Ultra2 SCSI, which all feature a 16-bit bus and from 40 to 80MBps transfer rate.

     All SCSI devices should be powered on before the PC to allow the SCSI host adapter to detect each of the devices on the SCSI bus.

RAID Technology

     A Redundant Array of Independant Disks (RAID) is a category of disk drives that employ two or more disks in combination for fault tolerance and performance. One of the fundamental concepts of RAID disk drives is data striping. In this process, data files are subdivided and written to several disks. This technology allows the processor to read or write data to the disk system faster than a single disk can supply or accept it. While the first data segment transfers from the first disk, the second disk is locating the next segment, and so on.

     Another common feature of RAID systems is data mirroring. This feature involves writing duplicate data segments or files to more than one disk to guard against losing the data due to device failure.

     Ten different RAID levels exist 0 - 7, 10, and 53, each more complicated than its predecessor. The most common are RAID 0, 3, and 5.

RAID 0 - Striped disk array without fault tolerance. This level provides for data striping but doesn't include mirroring or redundancy or any protection against device failure (fault tolerance).
RAID 3 - Parallel transfer with parity. This level is very much like level 0, except that it sets aside a dedicated disk for storing parity and error correction code (ECC) data.
RAID 5 - Data striping with parity. This level provides data striping at the character level and also implements stripe error correction. The ECC data is recorded on a seperate disk for each level of data stripe.

Working With Disk Storage

Following are the building blocks of disk media organization:

Tracks - Concentric circular areas of the disk. A length is one circumference of the disk. When data is written to the disk, it begins with the outermost track first.
Sectors - Cross-sectioning devisions of the disk that intersect all the tracks. In addition to dividing each track into manageable pieces, sectors provide addressing references.
Cylinders - Unique to hard disk drives, cylinders are a logical grouping of the same track on each disk surface. If a hard disk for four platters, it has eight surfaces and eight tracks of the same number. For example track 52 will exist on all eight sides and all eight track 52s make up cylinder 52. This feature allows data to be written to each platter on the same track, eliminating the need to move the read / write heads.
Clusters - Clusters are groups of sectors used by operating systems to track data on the disk. There are normally around 64 sectors to a cluster, but the size of the disk drive determines the actual number of sectors in a cluster.

Reading and Writing to A Disk

Before data can be read from a disk, it must be written to the disk through a process called flux transition, which means that the storage media is altered with a electromagnet to either a positive or a negative charge. Primarily, two different encoding schemes have been used to convert data into flux transition.

FM (frequency modulation) and MFM (modified frequency modulation) - Some of the first widely used encoding methods. These schemes simply recorded a 1 or 0 as different polarities on the recording media.
RLL(run length limited) - Allows for higher track and data density by spacing one-bits farther apart specially encoding each byte. RLL introduced data compression techniques, and most current disk drives (IDE and SCSI) use a form of RLL encoding.

     Interleaving data on a disk is a technique that allows the read / write head to use the rotation of the disk to its advantage. If a disk drive has an interleaving ration of 3:1 or 3-1, it writes one sector of data and then waits or skips two before writing the next, and so forth. If it has a ratio of 2:1 or 2-1=1, it writes every other sector (skips one). An interleaving of 1:1 is the same as having no interleaving at all.

Formatting The Disk

     A disk drive has two levels of formatting, the low-level format and the high-level format. Low-level formatting is performed at the manufacturer. Low-level formatting performs two major functions: It builds the File Allocation Table (FAT) file and records the location of all tracks and sectors on the device; The second function is that it physically scans the disk media for defects and records the location of any unusable areas found.

     A high-level format varies slightly between operating systems. A DOS format creates a sperate FAT table that's used by the operating system to track disk clusters.

Partitioning The Hard Disk

     Paritioning a hard disk is the process of dividing it into logical subdivisions, which are seen by the operating system as seperate logical (as opposed to physical) hard disks. Partioning also allows the use of different operating systems such as Linux which requires the EXT2 filesystem which is a different than the FAT filesystem. Only one filesystem can exist on a partition.

     Hard disks are divided into primary and extended partitions. Usually, the primary partition is the one used to boot the system. You can have up to 4 partitions on a hard disk. You can only have on extended partition. This means you can have a maximum of 4 primary partitions, or a 3 primary partitions and 1 extended partition. The extended partition can have logical partitions (up to 23) which must all exist on the extended partition.

     Partitioning disks can improve disk efficiency. Under DOS and Windows, cluster sizes are automatically assigned in proportion t the disk size. The bigger the disk, the bigger the clusters. Only one file can exist in a cluster (although they may span more than one) so any slack space between the end of the file and the end of the cluster is wasted. By seperating a large disks into smaller partitions, it will reduce the cluster sizes and eliminate wasted space.

     After a hard disk is partitioned, the first sector on cylinder 0 is reserved for the master boot record that contains the partition table. All partitions have a partition table, however, the master boot record partition table contains the mapping for all partitions on all drives. The master boot record uses the partition table to locate the active primary partition to boot the system.

Disk Compression

     If a disk is compressed, a disk compression utility must reside in memory and work between the operating system and the disk controller. These utilities do add some overhead to the process by slowing down file access from the compressed disk. If you boot from a floppy disk which does not load the compression utility, you will not be able to access the compressed disk.

     DOS and Windows 3.x use a routine called DBLSpace. Windows 95 uses a compression utility called DriveSpace which can compress data on removable media other than floppy disks, as well as floppy and hard disks. DriveSpace can compress drives of up to 512MB by creating a seperate uncompressed logical drive, called the host drive, where it stores CVF or Compressed Volume File, a form of VFAT for the compressed drive.

Backing Up Data

     You can create four different types of backups:

An archival backup (or full backup) is one that backups up the entire data contents of the hard disk.
An incremental backup contains only the files that have been modified since the last previous backup.
A differential backup copies all the data added or modified since the last full backup.
A copy backup is created by using a copy command to write a duplicate of a file, directory, or disk to another media.

Installing and Configuring Storage Devies

Floppy Drives

     Older 5 1/4 in. floppy drives and 3 1/2 in. floppy drives come in three package sizes (or form factors): full-height, half-height, and combination half-height. A full-height takes about two expansion slots and is common with older PCs. The half-height drives is one-half as tall as a full-height drives and is the size of one expansion slot on a PC case. The combination half-height form combines a 3.5 and 5.25 in. drive into one half-height device.

     The floppy drive cable is a 34-wire ribbon cable that is usually light blue in color with one edge painted painted either red or blue. The red or distinct edge goes to the 1 pin. Pin 1 is either marked with a 1 or a white dot on the conroller card connector. A floppy cable will usually have two two-connector sets for floppy drives. The first set (in the middle) are for the /dev/fd1 or B: drive. At the end of the cable, is the /dev/fd0 or A: drive connectors.

     The drive cable takes care of the configurations for the floppy drives. The drive jumbers are usually set to DS2, or drives select 2, at the factory. The twist in the cable near the end will !boolean the jumper settings for DS1.

     To absorb all signals and prevent signal echoes bouncing back down the cable and crashing into new incoming data, we need a terminating resistor plugged into the last drive on the cable. All 3.5 drives have a preinstalled terminating resistor so you don't have to have one for 3.5. If you mix a 5.25 drive, the terminator on the 5.25 should be removed unless it's at the end. The terminating resistors look like 16-pin memory DIP chips.

IDE/EIDE Drives

     IDE and EIDE drives are installed with a 40-pin ribbon cable that must also be aligned to pin 1. The alignment is the same as with floppy drives, the unique or red side is aligned to pin 1. Only two IDE drives can be installed in a system. One is the master the other is the slave. Up to four EIDE drives can be installed with two drives each on two cables; each cable should have a primary and slave designation.

SCSI Drives

     Each SCSI device must be assigned an ID number. It is set through a jumber or a DIP switch. SCSI can have 8 devices on the SCSI bus, one of which is device 7 is the SCSI host adapter. Slower devices such as CD-ROMs should be assigned a higher number to give them time to process data on the SCSI bus. A SCSI hard disk used to boot the computer is assigned SCSI ID 0.

If you have both an IDE and SCSI disk drive on a system, the IDE drive should be the boot drive.

A three-pin jumber can be used to indicate a total of eight values, 0-7.