RAIDBank4 Owner’s Manual
57
Appendix D- Glossary
ATA
Acronym for “
AT
Bus
A
ttachment” - a standard interface to IDE hard disks. Western
Digital’s IDE disk interface was standardized by ANSI to form the ATA specification using a
16-bit ISA bus.
Cache
cache is a fast-access memory bank that serves as an intermediate storage for data that
is read from or written to secondary storage. Typically, high-speed caches are implemented in
RAM, though they can also be implemented on disk when speed is not a critical requirement.
Caches generally improve the efficiency of read operations due to the principles of “spatial
and temporal locality of data”. They can also improve the efficiency of write operations.
See
also: Write Back Cache, Write Through Cache
Degraded Mode/Status
All arrays, with the exception of
RAID
0, are designed to handle
disk failures. However, there is limit on the number of hard disks that can fail before the
array is rendered inoperative. For instance, this limit value is 1 for
RAID
1, 3, and 5. In the
case of
RAID
10 or 50, the upper bound is equal to the number of parity groups. When the
number of disk failures occurring in an array are less than or equal to this upper bound, the
array is denoted to be in a degraded state. The failure of the disks does not impair reading
from or writing to the array. However, it impairs the efficiency of throughput in all RAID
types (with the exception of RAID 1) since data requested by read operations may have to be
“reconstructed” using parity. In the case of RAID 1 the throughput of read operations is cut in
half if a drive fails. Operating in degraded mode is considered an acceptable alternative only
for short durations. Generally this duration should span no more time than that required to
inform the user of the failures and to replace the failed disks with suitable spares.
Dirty Data
Dirty data is data that has been written to a cache but has not been “flushed,” or
written to its final destination, typically some secondary storage device.
Disk Array
A Disk Array is a logical disk comprised of multiple physical hard disks. The
number of hard disks in an disk array is dictated by the type of the array and the number of
spares that may be assigned to it. Furthermore, whether a disk array can be built using part
of the space on a disk (as opposed to being forced to use the whole disk) depends upon the
implementation. Disk Arrays are typically used to provide data redundancy and/or enhanced
I/O performance.
Disk Block
Data is stored on disks in blocks that are generally of a predefined size. This size
is typically a value such as 512 bytes, 1 KB, 2 KB, etc. When a record is written to a disk, the
blocks used for that record are dedicated to storing the data for that record only. In other
words two records are not permitted to share a block. Consequently, a block may be only
partially used. For instance, assume a disk has a block size of 1 KB and a user record written
to it has a size of 3148 bytes. This implies that the user record will be written into 4 blocks,
with the contents of one of the blocks being only partially filled with (3148 – 3072) 76 bytes
of data.
D-Glossary
