Data Protection and Information Lifecycle Management
< Day Day Up > |
Storage I/O Basics
Data storage uses a variety of technology to store data. Some devices such as hard drives, floppy drives, and tape use magnetism to encode information on the media. Others, such as CD-ROM/RW and DVD-ROM/RW, use lasers to burn information in the substrate of a disk. Still others, such as solid state storage, use a charge in a solid state device to store data. Whatever the media, all data is stored as bits (1s and 0s) on some device that can be read back later. To be found, stored data must have a structure that can be understood by an operating system. If this structure doesn't exist, data cannot be written to the storage media. Reading back data would be an exercise in futility. At the lowest level, the drive and protocol determine how data is organized and transferred to and from a device. Unfortunately, the high-level structure is imposed by the operating system. This makes file systems unique to the operating environment. It is why a file written to a hard drive by Microsoft Windows cannot be read by any other operating system, such as a Macintosh, without a translation program that understands both file systems.
How Data Is Accessed and Stored on Media
With the exception of solid state storage, data is stored by encoding a bit onto a material, called the media. The electromechanical devices that are used to read and write data are called the heads. The read and write heads are mounted on an arm called the head assembly or actuator assembly. Disks and tape write data by positioning the write heads over the place on the disk or tape where the data is to be stored. The head then changes the magnetic state at that spot. To read the information stored on the disk, the read head is positioned over the spot where the data is stored. For magnetic media, the head measures the magnetic field at that spot and determines whether a 1 or 0 is encoded there. For optical media, such as a CD-ROM or DVD-ROM, a laser is used to read and write data. To write, the laser burns a pit in the surface of the media to represent a 1. When reading, the laser shines on the surface and, depending on how it reflects back, determines whether a pit (a 1) is there. If a pit is not there, it is a 0. Access Time
The amount of time it takes for a controller to find and read data on a media is called access time. Access time is an important performance measure because it directly measures the time it takes to retrieve data from the media. With disk drives, access time is a measure of the latency and transfer time of the disk. Latency
Storage devices that are mechanical in nature take time to find the data on the media and place the read or write heads over it. The time it takes to place the heads over the spot where the data is stored or will be stored is called the latency. For hard disks, the read/write heads have to be positioned over the exact spot where the data has been stored or will be stored. The drive controller has to wait for the disk to spin around to the arc of the disk to the desired spot. It then extends the heads over the interior to that spot. All of this movement determines the latency of the disk drive. CD and DVD media works in a similar way, only using a laser instead of magnetic heads. A similar process takes place for magnetic tape. It is the magnetic tape that is moved, however, not the heads. A tape device's latency is the amount of time it takes to move the tape so that the tape heads are over the spot where data is to be written to or read from. Solid state media has incredibly low access time because it is made out of RAM. It does not suffer from the effects of slower mechanical components and, hence, has almost no latency. Transfer Time
Transfer time is the amount of time it takes the disk, tape, or other storage media to transfer data off the media and onto the data bus or network. It is a function of a number of variables, including how fast the heads can detect the magnetic field on the disk or tape (in the case of magnetic media), the performance of the drive mechanisms, and the speed of the electronic components. Latency and transfer time together help define the access time for disk-and tape-based media. Streaming Tape
Tape is different from disks. It streams, or moves continuously, from one end of the tape to the other unless told to stop. To find something on a digital tape (which is the prevailing type today), the tape drive controller reads information, encoded at the beginning of the tape, that tells it where the data is stored. It then begins to spool the tape forward until it reaches the desired position. Having previously read the data location information at the beginning of the tape, it can find additional blocks by moving the tape back and forth. This is inefficient, and a tape will wear out quickly if it is searched in a random fashion. Instead, tape software attempts to read and write starting at the beginning and not stopping until the very end of the data. Tape is best used in applications in which the tape can be kept streaming, such as backup and restore. |
< Day Day Up > |