Memory for disk arrays
The disadvantage of hardware implementation of RAID is the relatively high cost
of RAID controller. However, on the one hand, for all the (reliability,
performance, service) have to pay. On the other hand, more recently, the
development of microprocessor technology, the cost of RAID controller (especially
younger models) was to fall sharply and become comparable with the cost of
ordinary disk controllers, enabling RAID systems not only in plots to-mainframe,
but in the entry-level server and even workstations. Structurally controllers
are both the outside and inside. There are also integrated feature on the
motherboard RAID Controllers. In addition, the controllers differ by
derzhivaemym interface disks. For example, SCSI RAID controllers are designed
for use in servers, and IDE RAID controllers are suitable for entry-level
servers, as well as for working stations. A distinctive feature of RAID
controller is the number of supported channels to connect hard drives. Despite
the fact that one channel controller can be connected few SCSI disks with a
total capacity RAID array will be limited bandwidth of one channel, which
corresponds to the capacity SCSI interface. Thus, the use of multiple channels
can significantly increase the productivity disk subsystem tests. Using IDE RAID
controllers problem of multi-faced even greater, since two hard disks connected
to a single channel (more disks, not under-cism own interface), can not provide
parallel - IDE interface to allow contact at a particular time only one disk.
Therefore, IDE RAID controllers must be a minimum of two. There are also four -
and even eight-channel controller. Another difference between IDE RAID and SCSI
RAID controllers, the number of under-derzhivaemyh their levels. SCSI RAID
controllers support all major levels, and as the rule, and several more combined
company levels.
Next part
|