[ Team LiB ] Previous Section Next Section

Mass Storage Systems

Secondary storage is essentially an electronic closet, a place where you put information that you don't want to constantly hold in your hands but that you don't want to throw away, either. As with the straw hats, squash rackets, wallpaper tailings, and all the rest of your dimly remembered possessions that pile up out of sight behind the closet door, retrieving a particular item from mass storage can take longer than when you have what you want at hand.


Mass storage takes two forms: online storage, which is instantly accessible by your microprocessor's commands, and offline storage, which requires some extra, possibly human, intervention (such as you sliding a cartridge into a drive) for your system to get the bytes that it needs. Sometimes, the term near-line storage is used to refer to systems in which information isn't instantly available but can be put into instant reach by a microprocessor command. The jukebox—an automatic mechanism that selects CD-ROM cartridges, and sometimes tape cartridges—is the most common example.

Moving bytes from mass storage to memory determines how quickly stored information can be accessed. In practical online systems, the time required for this access ranges from less than 0.01 seconds in the fastest hard disks to 1000 seconds in some tape systems, spanning a range of 100,000 or five orders of magnitude.

By definition, the best offline storage systems have substantially longer access times than the quickest online systems. Even with fast-access disk cartridges, the minimum access time for offline data is measured in seconds because of the need to find and load a particular cartridge. The slowest online and the fastest offline storage system speeds, however, may overlap because the time to ready an offline cartridge can be substantially shorter than the period required to locate needed information written on a long online tape.

Data Organization

Computer mass storage systems differ in the way they organize and allow access to the information that they store. Engineers class the storage systems in modern computers as either sequential or random access. From a practical standpoint, the difference is functional—how the computer system finds the information it needs. But it also has a historic dimension—sequential computer mass storage predates random access. You can also look at the difference topographically. Sequential storage is one dimensional, whereas random access has two (or possibly more) dimensions.

Neither technology is inherently better. Both are used in modern computers. For a given application, of course, one of the two is typically more suitable. But because each has its own strengths, both will likely persist as long as computers populate desktops, if not longer.

Sequential Media

A fundamental characteristic of tape recording is that information is stored on tape one-dimensionally—in a straight line across the length of the tape. This form of storage is called sequential because all the bits of data are organized one after another in a strict sequence, like those paper-based dots. In digital systems, one bit follows after the other for the full length of the tape. Although in multiple-track and helical recording systems, bits may also spread across the width of the tape, conceptually these, too, store information in one dimension only.

CDs and DVDs designed for audio and video applications (that is, CD-DA, CD-Video, DVD-Video, and DVD-Audio) are designed to work best as one-dimensional media as well. They record data as a continuous stream meant to be displayed (or listened to) in the same sequential order it was recorded. However, the two-dimensional nature of the disc (a disc is a two-dimensional plane, after all) allows these media to transcend some of the limitations of sequential recording.

In the Newtonian universe (the only one that appears to make sense to the normal human mind), the shortest distance between two points is always a straight line. Alas, in magnetic tape systems, the shortest distance between two bits of data on a tape may also be a long time. To read two widely separated bits on a tape, all the tape between them must be passed over. Although all the bits in between are not to be used, they must be scanned in the journey from the first to second bits. If you want to retrieve information not stored in order on a tape, the tape must shuttle back and forth to find the data in the order that you want it. All that tape movement to find data means wasted time.

In theory, there's nothing wrong with sequential storage schemes—depending on the storage medium that's used, they can be very fast. For example, one form of solid-state computer memory, the all-electronic shift register, moves data sequentially at nearly the speed of light.

The sequential mass storage systems of today's computers are not so blessed with speed, however. Because of their mechanical foundations, most tape systems operate somewhat slower than the speed of light. For example, although light can zip across the vacuum of the universe at 186,000 miles per second (or so), cassette tape crawls along at one and seven-eighths inches per second. Although light can get from here to the moon and back in a few seconds, moving a cassette tape that distance would take about ten billion times longer, several thousand years.

Although no tape stretches as long as the 238,000 mile distance to the moon, sequential data access can be irritatingly slow. Instead of delivering the near-instant response most of today's impatient power users demand, picking a file from a tape can take as long as ten minutes. Even the best of today's tape systems require 30 seconds or more to find a file. If you had to load all your programs and data files from tape, you might as well take up crocheting to tide you through the times you're forced to wait.

Most sequential systems store data in blocks, sometimes called exactly that, and sometimes called records. The storage system defines the structure and contents of each block. Typically, each block includes identifying information (such as a block number) and error-control information in addition to the actual data. Blocks are stored in order on tape. In some systems, they lie end to end while others separate them with blank areas called inter-record gaps.

Most tape systems use multiple tracks to increase their storage (some systems spread as many as 144 tracks across tape just one-quarter inch wide). The otherwise stationary read-write head in the tape machine moves up and down to select the correct track. This multitrack design also improves the access speed of the medium. With multiple tracks, the drive doesn't have to scan the whole tape to get at a particular block. With 144 tracks, a drive needs only to scan 1/144th the length of the tape to find a particular block. Sequential CDs and DVDs provide a greater speed-up because they can select among thousands of radial positions along the disc to read.

To take advantage of the speed-up afforded by multiple tracks, the drive needs to know on which track to find the block you're seeking. Old tape systems didn't keep track of location information, so they would scan back and forth along the tape, one track after another, to find the blocks you wanted. New tape standards put a directory on the tape that holds the location of information on the tape. By consulting the directory, the drive can determine which track holds the information you want and zero in on the correct track to trim the response time of the tape system.

Random-Access Media

On floppy and most hard computer disks, as well as some CD and DVD formats, the recorded data is organized to take advantage of the two-dimensional aspect of the flat, wide disk surface to give even faster access than is possible with the directory system on tape. Instead of being arranged in a single straight line, disk-based data is spread across several concentric circles like lanes in a circular racetrack or the pattern of waves rolling away from a splash. Some optical drives follow this system, but many other optical systems modify this arrangement, changing the concentric circles into one tightly packed spiral that continuously winds from the edge to the center of the disk. But even these continuous-data systems behave much as if they had concentric circles of information.

The mechanism for making this arrangement is quite elementary. The disk moves in one dimension under the read/write head, which scans the tape in a circle as it spins and defines a track, which runs across the surface of the disk much like one of the lanes of a racetrack. In most disk systems, the head can move as well; otherwise, the read/write head would be stuck forever hovering over the same track and the same stored data, making it a sequential storage system that wastes most of the usable storage surface of the disk.

In most of today's disks systems, the read/write moves across a radius of the disk, perpendicular to a tangent of the tracks. The read/write head can quickly move between the different tracks on the disk. Although the shortest distance between two points (or two bytes) remains a straight line, to get from one byte to another, the read/write head can take shortcuts across the lanes of the racetrack. After the head reaches the correct track, it still must wait for the desired bit of information to cycle around under it. However, disks spin relatively quickly—300 revolutions per minute for most floppy disks and up to 7200 rpm for some hard disks—so you only need to wait a fraction of a second for the right byte to reach your system.

Because the head can jump from byte to byte at widely separated locations on the disk surface and because data can be read and retrieved in any order or at random in the two-dimensional disk system, disk storage systems are often called random-access devices, even though they fall a bit short of the mark with their need to wait while hovering over a track.

The random-access capability of magnetic disk systems makes the combination much faster than sequential tape media for the mass storage of data. Disks are so superior and so much more convenient than tapes that tape is almost never used as a primary mass storage system. Usually, tape plays only a secondary role as a backup system. Disks are used to store programs and files that need to be loaded on a moment's notice.

    [ Team LiB ] Previous Section Next Section